IJACSA Volume 14 Issue 11 - thesai.org

IJACSA Volume 14 Issue 11

Copyright Statement: This is an open access publication licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, even commercially as long as the original work is properly cited.

View Full Issue

Paper 1: Sentiment-Driven Forecasting LSTM Neural Networks for Stock Prediction-Case of China Bank Sector

Abstract: This study explores the predictive analysis of public sentiment in China's financial market, focusing on the banking sector, through the application of machine learning techniques. Specifically, it utilizes the Baidu Index and Long Short-Term Memory (LSTM) networks. The Baidu Index, akin to China's version of Google Trends, serves as a sentiment barometer, while LSTM networks excel in analyzing sequential data, making them apt for stock price forecasting. Our model integrates sentiment indices from Baidu with historical stock data of significant Chinese banks, aiming to unveil how digital sentiment influences stock price movements. The model's forecasting prowess is rigorously evaluated using metrics such as R-squared (R2), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and confusion matrices, the latter being instrumental in assessing the model's capability in correctly predicting stock up or down movements. Our findings predominantly showcase superior prediction performance of the sentiment-based LSTM model compared to a standard LSTM model. However, effectiveness varies across different banks, indicating that sentiment integration enhances prediction capabilities, yet individual stock characteristics significantly contribute to the prediction accuracy. This inquiry not only underscores the importance of integrating public sentiment in financial forecasting models but also provides a pioneering framework for leveraging digital sentiment in financial markets. Through this endeavor, we offer a robust analytical tool for investors, policymakers, and financial institutions, aiding in better navigation through the intricate financial market dynamics, thereby potentially leading to more informed decision-making in the digital age.

Author 1: Shangshang Jin

Keywords: Machine learning; LSTM; sentiment; forecasting; banking sector

Paper 2: Semantic Sampling: Enhancing Recommendation Diversity and User Engagement in the Headspace Meditation App

Abstract: In this paper, we present a clever approach to enhance the performance of sequential recommendation systems, specifically in the context of meditation recommendations within the Headspace app. Our method, termed “Semantic Sampling”, leverages the power of language embeddings and clustering techniques to introduce diversity and novelty in the recommen-dations. We augment the Time Interval Aware Self-Attention for Sequential Recommendation (TiSASRec) model with semantic sampling, where the next recommended item is randomly sampled from a cluster of semantically similar items. Our empirical evaluation, conducted on a sample set of 276,700 users, reveals a statistically significant increase of 2.26 % in content start rate for the treatment group (TiSASRec with semantic sampling) compared to the control group (TiSASRec alone). Furthermore, our approach demonstrates improved coverage and rarity, indi-cating a broader range of recommendations and higher novelty. The results underscore the potential of Semantic Sampling in enhancing user engagement and satisfaction in recommendation systems.

Author 1: Rohan Singh Rajput

Author 2: Christabelle Pabalan

Author 3: Akhil Chaturvedi

Author 4: Prathamesh Kulkarni

Author 5: Adam Brownell

Keywords: Information retrieval; machine learning; recom-mender system

Paper 3: State of the Art in Intent Detection and Slot Filling for Question Answering System: A Systematic Literature Review

Abstract: A Question Answering System (QAS), also known as a chatbot, is a Natural Language Processing (NLP) application that automatically provides accurate responses to questions posed by humans in natural language. Intent Detection and Classification are crucial elements in NLP, especially in a task-oriented dialogue system. In this paper, we conduct a systematic literature review that will perform a comparative analysis of different techniques or algorithms that are being implemented for intent detection and classification with slot filling. The goals of this paper are to identify the distribution, methodology, techniques or algorithms, and evaluation methods, that can be used to develop and construct a model of intent detection and classification with slot filling. This paper also reviews academic documents that have been published from 2019 to 2023, based on a four-step selection process of identification, screening, eligibility, and inclusion, for the selection process. In order to examine these documents, a systematic review was conducted and four main research questions were answered. The results discuss the methodology that can be used for the implementation of intent detection and classification with slot filling, along with the techniques, algorithms and evaluation methods that are widely used and currently implemented by other researchers.

Author 1: Anis Syafiqah Mat Zailan

Author 2: Noor Hasimah Ibrahim Teo

Author 3: Nur Atiqah Sia Abdullah

Author 4: Mike Joy

Keywords: Intent detection; intent classification; slot filling; question answering system

Paper 4: Enhancing Vehicle Safety: A Comprehensive Accident Detection and Alert System

Abstract: This research pioneers a ground-breaking system meticulously engineered to swiftly detect vehicular accidents and dispatch immediate alerts to both emergency services and pre-assigned contacts. This symphony of cutting-edge technologies includes an accelerometer sensor attuned to detect acceleration in any vector, a dynamic Liquid-Crystal Display (LCD) display for rapid alert dissemination, an assertive buzzer for resonant alarms, a Global System for Mobile (GSM) module for the swift transmission of distress messages, and pinpoint location data provided by a Global Positioning System (GPS) module. A user-friendly 'cancel' button acts as an escape hatch from potential false alarms. Orchestrated by the dexterity of an Arduino Uno microcontroller, this ensemble orchestrates a harmonious ballet of safety. This solution boasts cost-effectiveness, steadfastness, and unparalleled efficiency. Rigorous testing across diverse scenarios confirms its precision and robustness. By enhancing accident detection accuracy, expediting emergency responses, and facilitating rapid location dissemination, this innovation serves as a vital lifeline, empowering both passengers and rescue services upon accident initiation. With location data as its guiding star, emergency services gain a swift navigational edge, offering a beacon of hope in the battle against accident-related casualties.

Author 1: Jamil Abedalrahim Jamil Alsayaydeh

Author 2: Mohd Faizal bin Yusof

Author 3: Mohamad Amirul Aliff bin Abdillah

Author 4: Ahmed Jamal Abdullah Al-Gburi

Author 5: Safarudin Gazali Herawan

Author 6: Andrii Oliinyk

Keywords: Vehicle accident detection; microcontroller-based system; accelerometer sensor; Global Positioning System (GPS) localization; Global System for Mobile (GSM) communication; emergency response; safety innovation

Paper 5: Design of University Archives Business Data Push System Based on Big Data Mining Technology

Abstract: Aiming at the problems of low accuracy, recall, coverage and push efficiency of university archives business data, a university archives business data push system based on big data mining technology is designed. Firstly, the overall architecture and topological structure of the university archives business data push system are designed, and then the functional modules of the system are designed. Using big data mining technology to mine user behavior, modeling according to user behavior sequence, and designing a model to predict user behavior sequence based on hidden Markov model theory. Finally, the user behavior sequence is analyzed, and the factors such as user collaboration, similarity of user behavior sequence and data timeliness are comprehensively considered to push university archives business data for users. The experimental results show that the proposed method has high data push accuracy, recall, coverage and push efficiency, and can effectively push the required business data for users.

Author 1: Zhongke Wang

Author 2: Jun Li

Keywords: Big data mining technology; system design; business data pus; hidden markov model; similarity

Paper 6: Augmented Reality SDK Overview for General Application Use

Abstract: Augmented Reality Software Development Kits, or as they are commonly called AR SDKs, are useful for developers to build digital objects in AR. This paper presents a comparative study of AR SDKs. This comparison is based on several significant criteria, to select the most suitable SDK. The evaluation used the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) method. Based on a comparative analysis of the features and virtual elements available for application development with the AR SDK, researcher suggests that the main functions of the AR SDK were to be able to offer AR application Editing Platform and facilitate software creation without requiring knowledge of algorithms. Besides that, it is possible to establish some general observations regarding the benefits and limitations of the AR SDK. The result of this research is expected to provide with the clear framework for processing the data that has been collected, summarized, and tested from case study so the researcher will be able to reach useful conclusions. From the literature study has been conducted, it was concluded that among many SDK tools, there are 15 of them which were the most employed by AR developers. These 15 tools were selected based on certain main attributes and support platforms. At the end of this research, it also presents the advantages and limitations of these 15 tools.

Author 1: Suzanna

Author 2: Sasmoko

Author 3: Ford Lumban Gaol

Author 4: Tanty Oktavia

Keywords: Augmented reality; software development kits; AR SDK; platform; framework; AR technology

Paper 7: Automatic Extractive Summarization using GAN Boosted by DistilBERT Word Embedding and Transductive Learning

Abstract: Text summarization is crucial in diverse fields such as engineering and healthcare, greatly enhancing time and cost efficiency. This study introduces an innovative extractive text summarization approach utilizing a Generative Adversarial Network (GAN), Transductive Long Short-Term Memory (TLSTM), and DistilBERT word embedding. DistilBERT, a streamlined BERT variant, offers significant size reduction (approximately 40%), while maintaining 97% of language comprehension capabilities and achieving a 60% speed increase. These benefits are realized through knowledge distillation during pre-training. Our methodology uses GANs, consisting of the generator and discriminator networks, built primarily using TLSTM - an expert at decoding temporal nuances in timeseries prediction. For more effective model fitting, transductive learning is employed, assigning higher weights to samples nearer to the test point. The generator evaluates the probability of each sentence for inclusion in the summary, and the discriminator critically examines the generated summary. This reciprocal relationship fosters a dynamic iterative process, generating top-tier summaries. To train the discriminator efficiently, a unique loss function is proposed, incorporating multiple factors such as the generator’s output, actual document summaries, and artificially created summaries. This strategy motivates the generator to experiment with diverse sentence combinations, generating summaries that meet high-quality and coherence standards. Our model’s effectiveness was tested on the widely accepted CNN/Daily Mail dataset, a benchmark for summarization tasks. According to the ROUGE metric, our experiments demonstrate that our model outperforms existing models in terms of summarization quality and efficiency.

Author 1: Dongliang Li

Author 2: Youyou Li

Author 3: Zhigang ZHANG

Keywords: Extractive text summarization; generative adversarial network; transductive learning; long short-term memory; DistilBERT

Paper 8: Security in Software-Defined Networks Against Denial-of-Service Attacks Based on Increased Load Balancing Efficiency

Abstract: The goal of software-oriented networks (SDNs), which enable centralized control by separating the control layer from the data layer, is to increase manageability and network compatibility. However, this form of network is vulnerable to the control layer going down in the face of a denial-of-service assault because of the centralized control policy. The considerable increase in events brought on by the introduction of fresh currents into the network puts a lot of strain on the control surface when the system is in reaction mode. Additionally, the existence of recurring events that seriously impair the control surface's ability to function, such as the gathering of statistical data from the entire network, might have a negative impact. This article introduces a new approach that uses a control box comprising a coordinating controller, a main controller that establishes the flow rules, and one or more sub-controllers that establish the rules to fend off the attack and avoid network paralysis. It makes use of current (when needed). The controllers who currently set the regulations are relieved of some work by giving the coordinating controller management and supervision responsibilities. Additionally, the coordinator controller distributes the load at the control level by splitting up incoming traffic among the controllers of the flow rules. Thus, a proposed method can avoid performance disruption of the flow rule setter's main controller and withstand denial-of-service attacks by distributing the traffic load brought on by the denial-of-service attack to one or more sub-controllers of the flow rule setter. The results of the experiments conducted indicate that, when compared to the existing solutions, the proposed solution performs better in the face of a denial-of-service assault.

Author 1: Ying ZHANG

Author 2: Hongwei DING

Keywords: Security; open balance; denial-of-service attacks; software-oriented networks

Paper 9: Optimization of Unsupervised Neural Machine Translation Based on Syntactic Knowledge Improvement

Abstract: Unsupervised Neural Machine Translation is a crucial machine translation method that can translate in the absence of a parallel corpus and opens up new avenues for intercultural dialogue. Existing unsupervised neural machine translation models still struggle to deal with intricate grammatical relationships and linguistic structures, which leads to less-than-ideal translation quality. This study combines the Transformer structure and syntactic knowledge to create a new unsupervised neural machine translation model, which enhances the performance of the existing model. The study creates a neural machine translation model based on the Transformer structure first, and then introduces sentence syntactic structure and various syntactic fusion techniques, also known as the Transformer combines grammatical knowledge. The results show that the Transformer combines grammatical knowledge paired with Bi-Long Short-Term Memory proposed in this research has better performance. The accuracy and F1 value of the combined model in the training dataset are as high as 0.97. In addition, the time of the model in real sentence translation is controlled within 2s, and the translation accuracy is above 0.9. In conclusion, the unsupervised neural machine translation model proposed in this study has better performance, and its application to actual translation can achieve better translation results.

Author 1: Aiping Zhou

Keywords: Unsupervised; Neural network; Machine translation; Grammatical knowledge; Transformer; LSTM

Paper 10: Construction of a Security Defense Model for the University's Cyberspace Based on Machine Learning

Abstract: In order to ensure the security of university teachers and students using cyberspace, a machine learning based university cyberspace security defense model is constructed. Adopting a compression perception based data collection method for university cyberspace, the data information collection of university cyberspace is completed through sparse representation, compression measurement, and recovery reconstruction. Combining the advantages of Convolutional Neural Network (CNN) model in spatial feature extraction of data and Long Short Term Memory (LSTM) model in sequential feature extraction of data, extract the features of university network spatial data. After completing the multi feature dimensionality reduction processing of university network data based on the non-negative matrix decomposition algorithm, the feature dimensionality reduction processing results are input into the ConvLSTM-CNN model. After convolution calculation and integration, the security threat detection results of university network space are output. Based on the results of security threat detection, corresponding network attack defense measures are selected to ensure the security of the university's cyberspace. The experimental results show that the average attack interception rate of the model after application can reach 97.6%. It has been proven that building a model can accurately detect security threats to the university's cyberspace and achieve defense against various network attacks in different environments.

Author 1: Wang Bin

Keywords: Machine learning; University's cyberspace; security defense; construction of a model; compressed sensing; non-negative matrix

Paper 11: Investigating Efficiency of Soil Classification System using Neural Network Models

Abstract: Soil is a vital requirement for agricultural activities providing numerous functionalities restoring both abiotic and biotic materials. There are different types of soils, and each type of soil possesses distinctive characteristics and unique harvesting properties that impact agricultural development in various ways. Generally, farmers in the olden days used to analyse soil by looking at it visually while some prefer laboratory tests which are time-consuming and costly. Testing of soil is done to analyse the features and characteristics of the soil type, which results in selecting a suitable crop. This in turn results in increased food productivity which is very beneficial to farmers. Hence, to recognize the soil type an automatic soil identification model is proposed by implementing Deep Learning Techniques. It is used to classify the soil for crop recommendation by analysing accurate soil type. Different Convolution Neural Networks have been applied in the proposed model. They are VGG16, VGG19, InceptionV3 and ResNet50.Among all those techniques it is analysed that better results were obtained with ResNet50 having an accuracy of about 87% performing Multi-classification that is Black soil, Laterite Soil, Yellow Soil, Cinder soil & Peat soil.

Author 1: Pappala Mohan Rao

Author 2: Kunjam Nageswara Rao

Author 3: Sitaratnam Gokuruboyina

Author 4: Neeli Koti Siva Sai Priyanka

Keywords: Agricultural; convolution neural network; soil classification deep learning; VGG16; VGG19; InceptionV3; multi-classification; ResNet50

Paper 12: An Empirical Study: Automating e-Commerce Product Rating Through an Analysis of Customer Review

Abstract: e-Commerce today is a remarkable experience. However, finding and purchasing a right quality product based on numerous product reviews and manual rating in the e-commerce websites utilize much time among the consumers. This paper presents the problems faced by the consumers when buying products in e-commerce websites and a solution to solve the problems. Thus, the idea of an automated product rating system would be very useful for the consumers in which it rates the products automatically based on the reviews given by the buyers. To do this, a technique called Sentiment Analysis is used. It also ranks the products in order based on the product rating that is generated automatically. It would provide a way for the consumers to purchase their desired product within minutes. Surveys and interviews were conducted to find out the problems faced by consumers when purchasing a product online through e-commerce websites. There was also research conducted to study the product rating and product review section on the current e-commerce websites. To conclude, this automated product rating system eventually eases the consumers’ effort and time from reading numerous reviews and trusting inaccurate product rating to find a best quality product for them.

Author 1: Uvaaneswary Rajendran

Author 2: Salfarina Abdullah

Author 3: Khairi Azhar Aziz

Author 4: Sazly Anuar

Keywords: e-commerce website; sentiment analysis technique; manual product rating; automated product rating; product review

Paper 13: Secure IoT Routing through Manifold Criterion Trust Evaluation using Ant Colony Optimization

Abstract: The paper presents a simplified yet innovative computational framework to enable secure routing for sensors within a vast and dynamic Internet of Things (IoT) environment. In the proposed design methodology, a unique trust evaluation scheme utilizing a modified version of Ant Colony Optimization (ACO) is introduced. This scheme formulates a manifold criterion for secure data transmission, optimizing the sensor's residual energy and trust score. A distinctive pheromone management is devised using trust score and residual energy. Concurrently, several attributes are employed for constraint modeling to determine a secure data transmission path among the IoT sensors. Moreover, the trust model introduces a dual-tiered system of primary and secondary trust evaluations, enhancing reliability towards securing trusted nodes and alleviating trust-based discrepancies. The comprehensive implementation of the proposed integrates mathematical modeling, leveraging a streamlined bioinspired approach of the revised ACO using crowding distance. Quantitative results demonstrate that our approach yields a 35% improvement in throughput, an 89% reduction in delay, a 54% decrease in energy consumption, and a 73% enhancement in processing speed compared to prevailing secure routing protocols. Additionally, the model introduces an efficient asynchronous updating rule for local and global pheromones, ensuring greater trust in secure data propagation in IoT.

Author 1: Afsah Sharmin

Author 2: Rashidah Funke Olanrewaju

Author 3: Burhan Ul Islam Khan

Author 4: Farhat Anwar

Author 5: S. M. A. Motakabber

Author 6: Nur Fatin Liyana Mohd Rosely

Author 7: Aisha Hassan Abdalla Hashim

Keywords: Internet of things (IoT); secure IoT routing; manifold criterion trust evaluation; ant colony optimization (ACO); bioinspired computing; pheromone management

Paper 14: Analyzing Sentiment in Terms of Online Feedback on Top of Users' Experiences

Abstract: Since most businesses today are conducted online, it is crucial that each customer provide feedback on the various items offered. Evaluating online product sentiment and making suggestions using state-of-the-art machine learning and deep learning algorithms requires a comprehensive pipeline. Thus, this paper addresses the need for a comprehensive pipeline to analyze online product sentiment and recommend products using advanced machine learning and deep learning algorithms. The methodology of the research is divided into two parts: the Sentiment Analysis Approach and the Product Recommendation Approach. The study applies several state-of-the-art algorithms, including Naïve Bayes, Logistic Regression, Support Vector Machine (SVM), Decision Tree, Random Forest, Bidirectional Long-Short-Term-Memory (BI-LSTM), Convolutional Neural Network (CNN), Long-Short-Term-Memory (LSTM), and Stacked LSTM, with proper hyperparameter optimization techniques. The study also uses the collaborative filtering approach with the k-Nearest Neighbours (KNN) model to recommend products. Among these models, Random Forest achieved the highest accuracy of 95%, while the LSTM model scored 79%. The proposed model is evaluated using Receiver Operating Characteristic (ROC) - Area under the ROC Curve (AUC). Additionally, the study conducted exploratory data analysis, including Bundle or Bought-Together analysis, point of interest-based analysis, and sentiment analysis on reviews (1996-2018). Overall, the study achieves its objectives and proposes an adaptable solution for real-life scenarios.

Author 1: Mohammed Alonazi

Keywords: Sentiment analysis; product review; machine learning; recommendation system; collaborative filtering; exploratory data analysis

Paper 15: Automatic Model for Postpartum Depression Identification using Deep Reinforcement Learning and Differential Evolution Algorithm

Abstract: Postpartum depression (PPD) affects approximately 12% of new mothers, posing a significant health concern for both the mother and child. However, many women with PPD do not receive proper care. Preventative interventions are more cost-effective for high-risk women, but identifying those at risk can be challenging. To address this problem, we present an automatic model for PPD using a deep reinforcement learning approach and a differential evolution (DE) algorithm for weight initialization. DE is known for its ability to search for global optima in high-dimensional spaces, making it a promising approach for weight initialization. The policy of the model is based on an artificial neural network (ANN), treating the categorization issue as a policymaking stage-by-stage process. The DE algorithm is used to acquire initial weight values, with the agent obtaining samples and performing classifications in each step. The habitat provides an award for every categorization activity, considering a greater award for identification of the minor category to encourage precise detection. By using a particular compensatory technique and an encouraging learning system, the operator eventually decides the most excellent method for achieving its goals. The model's efficiency is evaluated by analyzing a set of data acquired from the population-based BASIC study carried out in Uppsala, Sweden, which covers the period from 2009 to 2018 and consists of 4313 samples. The experiential results, identified by known analysis criteria, indicate that the sample achieved better precision and correctness, making it suitable for identifying PPD. The proposed model could have significant implications for identifying at-risk women and providing timely interventions to improve maternal and child health outcomes.

Author 1: Sunyuan Shen

Author 2: Sheng Qi

Author 3: Hongfei Luo

Keywords: Postpartum depression; deep reinforcement learning; differential evolution algorithm; weight initialization; artificial neural network

Paper 16: Secure Cloud-Connected Robot Control using Private Blockchain

Abstract: With the increasing demand for remote operations and the challenges posed during the COVID-19 pandemic, industries across various sectors, including logistics, manufacturing, and education, have adopted virtual solutions. Cloud-based robot control has emerged as a viable approach for enabling safe remote operation of robots. However, along with the benefits, there are also risks associated with cloud-based robots. In this study, a secure cloud-based robot control system using a blockchain system was developed. The robot utilizes supervisory control to navigate via the internet. The communication system of the robot relies on the ThingsSentral cloud-based IoT platform; enabling communication between the user via a GUI developed using Python Tkinter and the local robot over the internet. To facilitate internet communication, the robot in this study incorporates an ESP32 microcontroller, which provides a low-cost and low-power system capable of connecting to Wi-Fi. However, cloud-based control systems are susceptible to cyberattacks, prompting the use of blockchain cybersecurity in this study to mitigate the risks. The data sent by the supervisor is stored within a private blockchain developed using Python, simultaneously being transmitted to the cloud platform. The developed security system addresses the risks associated with cloud-based robot control systems, such as data tampering and unauthorized misuse, by leveraging the Proof of Work (PoW) and hashing mechanisms.

Author 1: Muhammad Amzie Muhammad Fauzi

Author 2: Mohamad Hanif Md Saad

Author 3: Sallehuddin Mohamed Haris

Author 4: Marizuana Mat Daud

Keywords: Internet of Things (IoT); robot control; cloud computing; cybersecurity; blockchain

Paper 17: An Edge Computing-based Handgun and Knife Detection Method in IoT Video Surveillance Systems

Abstract: Real-time handgun and knife detection on edge devices within the Internet of Things (IoT) video surveillance systems hold paramount importance in ensuring public safety and security. Numerous methods have been explored for handgun and knife detection in video-based surveillance systems, with deep learning-based approaches demonstrating superior accuracy compared to other methods. However, the current research challenge lies in achieving high accuracy rates while managing the computational demands to meet real-time requirements. This paper proposes a solution by introducing a single-stage convolutional neural network (CNN) model tailored to address this challenge. The proposed method is developed using a custom dataset, encompassing model generation, training, validation, and testing phases. Extensive experiments and performance evaluations substantiate the efficacy of the proposed approach, which achieves remarkable accuracy results, thus showcasing its potential for enhancing real-time handgun and knife and knife detection capabilities in IoT-based video surveillance systems.

Author 1: Haibo Liu

Author 2: Zhubing HU

Keywords: Real-time detection; handgun and knife detection; edge devices; IoT video surveillance; deep learning; convolutional neural network

Paper 18: Advanced Seismic Magnitude Classification Through Convolutional and Reinforcement Learning Techniques

Abstract: Earthquake Early Warning (EEW) systems are crucial in reducing the dangers associated with earthquakes. This paper delves into the realm of EEWs, focusing on rapidly determining earthquake magnitudes (EMs). Traditional methods for swift magnitude categorization often grapple with challenges such as data disparity and cumbersome processes. Our research introduces an innovative EEW model, employing a 7-second seismic waveform record from three different components provided by the China Earthquake Network Center (CENC). This empirical, quantitative study pioneers a method combining dilated convolutional techniques with a novel mutual learning-based artificial bee colony (ML-ABC) algorithm and reinforcement learning (RL) for EM classification. The proposed model utilizes an ensemble of convolutional neural networks (CNNs) to simultaneously extract feature vectors from input images, which are then amalgamated for classification. To address the imbalances in the dataset, we implement an RL-based algorithm, conceptualizing the training process as a series of decisions with individual samples representing distinct states. Within this framework, the network operates as an agent, receiving rewards or penalties based on its precision in distinguishing between the minority and majority classes. A key innovation in our approach is the initial weight pre-training using the ML-ABC method. This technique dynamically optimizes the "food source" for candidates, integrating mutual learning elements related to the initial weights. Extensive experiments were carried out on the selected dataset to ascertain the most effective parameter values, including the reward function. The findings demonstrate the superiority of our proposed model over other evaluated methods, highlighting its potential as a robust tool for EM classification in seismology. This research provides valuable insights for both seismologists and developers of EEW systems, offering a novel, efficient approach to earthquake magnitude determination.

Author 1: Qiuyi Lin

Author 2: Jin Li

Keywords: Earthquake early warning; the magnitude of the earthquake; imbalanced classification; artificial bee colony; reinforcement learning

Paper 19: Information Retrieval System for Scientific Publications of Lampung University by using VSM, K-Means, and LSA

Abstract: The Lampung University repository system is a repository of data related to study, community service, and other scientific works, currently has 37242 documents accessible through repository.lppm.unila.ac.id. Despite the amount of data, its optimal use as an information retrieval remains unrealized, hindering the effective promotion of Lampung University's scientific publication excellence. Recognizing the limitations of existing information retrieval systems that are limited to specific methods for topic identification through clustering, this study aims to develop a retrieval system for Lampung University's repository using Vector Space Model (VSM), K-Means and Latent Semantic Analysis (LSA) that generates clusters and study expertise at the level of study program, faculty and Lampung University. The methodology includes data collection, preprocessing, modeling, evaluation and system deployment. The results show that the number of clusters obtained for the university level is 7 clusters, for the faculty level are 6, 7, 8 and 10 clusters, and for the program level are 3 to 5 clusters. In addition, the finding topic identification indicate that the expertise topics at Lampung University, which are agriculture, soil, education, plants, learning, society, Lampung. This study contributes to optimizing the information retrieval system, promoting academic excellence, and advancing the understanding of study expertise at Lampung University.

Author 1: Rahman Taufik

Author 2: Didik Kurniawan

Author 3: Anie Rose Irawati

Author 4: Dewi Asiah Shofiana

Keywords: Information retrieval; Vector Space Model (VSM); k-means; Latent Semantic Analysis (LSA); clustering; topic identification; scientific publication information

Paper 20: Adaptive Gray Wolf Optimization Algorithm based on Gompertz Inertia Weight Strategy

Abstract: To solve the problems that the Gray Wolf Optimizer (GWO) convergence speed is not fast enough and the solution accuracy is not high enough, this paper proposes an Adaptive Gray Wolf Optimizer based on Gompertz inertia weighting strategy (GGWO). GGWO uses the characteristics of the Gompertz function to achieve nonlinear adjustment of the inertia weight, which better balances the speed of global search and accuracy of local search of the GWO algorithm. At the same time, the Gompertz function is used to realize the adaptive adjustment of the individual gray wolf’s position and to better update the gray wolves’ position according to the fitness values of different gray wolf individuals. Use 6 classic test functions to compare the performance of GGWO in optimization and 10 other classic or improved swarm intelligence algorithms. Results show that GGWO has better solution accuracy, stability, and faster convergence than all other 10 swarm intelligence algorithms.

Author 1: Qiuhua Pan

Keywords: Gray wolf optimization algorithm; inertia weight; adaptive; Gompertz function; swarm intelligence algorithm

Paper 21: Optimizing Shuttle-Bus Systems in Mega-Events using Computer Modeling: A Case Study of Pilgrims' Transportation System

Abstract: Mega-events are held in a city, or more, during a limited time, which requires special attention to the infrastructure and the offered services. The Hajj event, hosted in Makkah - Saudi Arabia, is considered as an excellent example of religious mega-events. The field of computer modeling and simulation is one of the main technical tools that help in developing and understanding the risks of crowds and studying the safety means during organizing of many major events in the world. This paper focuses on using computer simulation to optimize the pilgrims' shuttle-bus transportation system in Holy Sites (Mashaaer), as a case study of optimizing shuttle-bus Systems in Mega-Events using computer modeling. The objective of paper is to develop a model of the shuttle-bus transport system to give insights of the advantages of the use as an alternative for transporting pilgrims as well as to provide decision makers with a tool that could be used to select the best parameters of the system for the most efficient operation. For this purpose, pilgrims’ evacuation time, traffic congestion and average trip time, from Arafat to Muzdalifa, are identified as the performance measures for evaluating the proposed transport system. The conducted simulation can be used to assess the current systems, recommend changes to the systems, and offer indicators and readings to assist decision makers.

Author 1: Mohamed S. Yasein

Author 2: Esam Ali Khan

Keywords: Computer modeling; simulation; optimization; shuttle-bus systems; mega-events; hajj

Paper 22: Development of Nursing Process Expert System for Android-based Nursing Student Learning

Abstract: Nurses are professionals who provide health services using a scientific process called nursing. In nursing, problem-solving uses the nursing process which is a critical thinking method, nurses must analyze the data found in patients to diagnose and determine the results and appropriate intervention plans. Prospective nursing students are required to be able to apply the nursing process in carrying out nursing care according to existing nursing standards, of course, with supervision by nursing experts to improve the quality of medical services. This study aims to develop an android application with the help of an expert system as a nursing diagnosis tool, which helps nursing students learn the nursing process and helps lecturers monitor the nursing process carried out by nursing students. This research uses 116 symptom data, 22 diagnosis data, 60 intervention data, 8 type data, and 864 description data. The results of this research are in the form of an expert system with an android-based forward chaining method that has been tested using the black box testing method.

Author 1: Aristoteles

Author 2: Abie Perdana Kusuma

Author 3: Anie Rose Irawati

Author 4: Dwi Sakethi

Author 5: Lisa Suarni

Author 6: Dedy Miswar

Author 7: Rika Ningtias Azhari

Keywords: Classification; expert system; forward chaining; blackbox testing; android; flutter; nursing process

Paper 23: Emotional State Prediction Based on EEG Signals using Ensemble Methods

Abstract: The emotional state is an essential factor that affects mental health. Electroencephalography (EEG) signal analysis is a promising method for detecting emotional states. Although multiple studies exist on EEG emotional signals classification, they have rarely considered processing time as a metric for classification model evaluation. Instead, they used either model accuracy and/or the number of features for evaluation. Processing time is an important factor to be considered in the context of mental health. Many people commonly use smart devices, such as smartwatches to monitor their emotional state and such devices require a short processing time. This research proposes an EEG-based model that detects emotional signals based on three factors: accuracy, number of features, and processing time. Two feature extraction algorithms were applied to EEG emotional signals: principal components analysis (PCA) and fast independent components analysis (FastICA). In the classification process, ensemble method classifiers were adopted due to their powerful performance. Three ensemble classifiers were used: random forest (RF), extreme gradient boosting (XGBoost), and adaptive boost (AdaBoost). The experimental results showed that the RF and XGBoost achieved the best accuracy, i.e., 95%, for both methods. However, XGBoost outperformed RF in terms of the number of features; it used 33 components extracted by PCA within 14 seconds, while RF used 36 within 4 seconds. AdaBoost was the worst in terms of both accuracy and processing time in the two experiments.

Author 1: Norah Alrebdi

Author 2: Amal A. Al-Shargabi

Keywords: Electroencephalograph; mental health; feature extraction; random forest; extreme gradient boosting; adaptive boost

Paper 24: Selection of a Trustworthy Technique for Fraud Prevention in the Digital Banking Sector

Abstract: Digital banking fraud poses a threat to the global economy and fintech applications. Sustainable models are essential to address this issue and minimize its economic impact. Hybrid methods have been developed to assess strategies for preventing digital banking fraud, aiding global stakeholders in making well-informed judgments. However, many of these models concentrate on the numerical features of digital banking ratios while overlooking crucial financial fraud protection qualities. This paper introduces a computational method for discovering and measuring the influence of digital banking fraud prevention strategies on sustainable fraud prevention. This innovative approach combines intuitionistic fuzzy set theory and the analytical network process for decision-making. Initially, an intuitionistic fuzzy expert system prioritizes crucial indices based on the preferences of financial decision-makers. This technique is then compared to alternative decision-making models across multiple variables. Empirical data demonstrate the superiority of the intuitionistic fuzzy-based decision-making system, outperforming other models and facilitating the recognition of financial statement fraud in global banking networks. Consequently, it offers a sustainable fintech solution. The findings of this study are pertinent to fintech scholars and practitioners engaged in the global battle against digital banking fraud.

Author 1: Bandar Ali M. Al-Rami Al-Ghamdi

Keywords: Digital banking fraud; analytical network process; intuitionistic fuzzy sets; fraud prevention and detection

Paper 25: Arabic Regional Dialect Identification (ARDI) using Pair of Continuous Bag-of-Words and Data Augmentation

Abstract: Author profiling is the process of finding characteristics that make up an author’s profile. This paper presents a machine learning-based author profiling model for Arabic users, considering the author’s regional dialect as a crucial characteristic. Various classification algorithms have been implemented: decision tree, KNN, multilayer perceptron, random forest, and support vector machines. A pair of Continuous Bag-of-Word (CBOW) models has been used for word representation. A well-known data set has been used to evaluate the proposed model and a data augmentation process has been implemented to improve the quality of training data. Support vector machines achieved a 50.52% f1-score, outperforming other models.

Author 1: Ahmed H. AbuElAtta

Author 2: Mahmoud Sobhy

Author 3: Ahmed A. El-Sawy

Author 4: Hamada Nayel

Keywords: Dialect identification; continuous Bag-of-Words; data augmentation; text classification.

Paper 26: Advanced Metering Infrastructure Data Aggregation Scheme Based on Blockchain

Abstract: Smart grid stands as both the cornerstone of the modern energy system and the pivotal technology for addressing energy-related challenges. Advanced Metering Infrastructure constitute a critical component within the smart grid ecosystem, providing real-time energy consumption data to power utility companies. Advanced Metering Infrastructure enables these companies to make timely and accurate decisions. Hence, the issue of data security pertaining to Advanced Metering Infrastructure assumes profound significance. Presently, Advanced Metering Infrastructure data confronts challenges associated with centralized data storage, rendering it susceptible to potential cyberattacks. Moreover, with the burgeoning number of electricity consumers, the resultant data volumes have swelled considerably. Consequently, the transmission of this data becomes intricate and its efficiency is compromised. To address these issues, this paper presents a lightweight blockchain data aggregation scheme. By integrating fog computing and cloud computing, a three-tier blockchain-based architecture is devised. Initially, digital signatures are employed to ensure the validity and integrity of user data. The innate attributes of blockchain technology are harnessed to safeguard the security of electricity energy data. Through secondary data aggregation, the privacy-sensitive user data is efficiently compressed and subsequently integrated into the blockchain, thereby mitigating the storage pressure on the blockchain and enhancing data transmission efficiency. Ultimately, through rigorous theoretical analysis and simulated experimentation, the paper demonstrates that, in comparison to existing methodologies, lightweight blockchain data aggregation scheme exhibits heightened security. Additionally, lightweight blockchain data aggregation scheme holds a competitive advantage in terms of computational and communication costs.

Author 1: Hongliang TIAN

Author 2: Naiqian ZHENG

Author 3: Yuzhi JIAN

Keywords: Smart grid; blockchain; advanced metering infrastructure; data aggregation

Paper 27: Predicting and Improving Behavioural Factors that Boosts Learning Abilities in Post-Pandemic Times using AI Techniques

Abstract: Quantifying student academic performance has always been challenging as it hinges on several factors including academic progress, personal characteristics and behaviours relating to learning activities. Several research studies are therefore being conducted to identify the factors so that appropriate measures can be conducted by academic institutions, family and the student to boost his/ her academic performance. The present study investigates personal characteristics, psychological factors, behavioural factors, social factors and learning capabilities, that directly or indirectly affect student’s academic performance, which was tapped by administering a self-designed questionnaire. The data was collected from 214 undergraduate students studying in various streams of the University of Delhi and post that semi-structured interview was conducted to get in- depth information. The result proved the correlation between the aforementioned factors and the learning capabilities of the students. Using the results of analysis a machine learning model based on k-nn algorithm was formed to predict student performance. A chatbot is also proposed to provide guidance to students in strenuous situations, motivate them and interact with them without having personal bias.

Author 1: Jaya Gera

Author 2: Ekta Bhambri Marwaha

Author 3: Reema Thareja

Author 4: Aruna Jain

Keywords: Academic performance; machine learning; chatbot; educational data mining; learning analytics

Paper 28: Sentiment Analysis Predictions in Digital Media Content using NLP Techniques

Abstract: In the current digital landscape, understanding sentiment in digital media is crucial for informed decision-making and content quality. The primary objective is to improve decision-making processes and enhance content quality within this dynamic environment. To achieve this, a comprehensive comparative analysis of NLP for tweet sentiment analysis was conducted, revealing compelling insights. The BERT pre-trained model stood out, achieving an accuracy rate of 94.56%, emphasizing the effectiveness of transfer learning in text classification. Among machine learning algorithms, the Random Forest model excelled with an accuracy rate of 70.82%, while the K Nearest Neighbours model trailed at 55.36%. Additionally, the LSTM model demonstrated excellence in Recall, Precision, and F1 metrics, recording values of 81.12%, 82.32%, and 80.12%, respectively. Future research directions include optimizing model architecture, exploring alternative deep learning approaches, and expanding datasets for improved generalizability. While valuable insights are provided by our study, it is important to acknowledge its limitations, including a Twitter-centric focus, constrained model comparisons, and binary sentiment analysis. These constraints highlight opportunities for more nuanced and diverse sentiment analysis within the digital media landscape.

Author 1: Abdulrahman Radaideh

Author 2: Fikri Dweiri

Keywords: Sentiment analysis; digital media; decision-making; quality assurance; NLP

Paper 29: An Enhanced Approach for Realizing Robust Security and Isolation in Virtualized Environments

Abstract: Transitioning into the next generation of supercomputing resources, we’re faced with expanding user bases and diverse workloads, increasing the demand for improved security measures and deeper software compartmentalization. This is especially pertinent for virtualization, a key cloud computing component that’s at risk from attacks due to hypervisors’ integration into privileged OSs and shared use across VMs. In response to these challenges, our paper presents a two- pronged approach: introducing secure computing capabilities into the HPC software stack and proposing SecFortress an enhanced hypervisor design. By porting the Kitten Lightweight Kernel to the ARM64 architecture and integrating it with the Hafnium hypervisor, we substitute the Linux-based resource management infrastructure, reducing overheads. Concurrently, SecFortress employs a nested kernel approach, preventing outerOS from accessing mediator’s memory, and creating a hypervisor box to isolate untrusted VMs’ effects. Our initial results highlight significant performance improvements on small scale ARM-based SOC platforms and enhanced hypervisor security with minimal runtime overhead, establishing a solid foundation for further research in secure, scalable high- performance computing.

Author 1: Rawan Abuleil

Author 2: Samer Murrar

Author 3: Mohammad Shkoukani

Keywords: Virtual Machine (VM); High-Performance Computing (HPC); cybersecurity; hypervisor security

Paper 30: Efficient Evaluation of SLAM Methods and Integration of Human Detection with YOLO Based on Multiple Optimization in ROS2

Abstract: In the realm of robotics, indoor robotics is an increasingly prominent field, and enhancing robot performance stands out as a crucial concern. This research undertakes a comparative analysis of various Simultaneous Localization and Mapping (SLAM) algorithms with the overarching objective of augmenting the navigational capabilities of robots. This is accomplished within an open-source framework known as the Robotic Operating System (ROS2) in conjunction with additional software components such as RVIZ and Gazebo. The central aim of this study is to identify the most efficient SLAM approach by evaluating map accuracy and the time it takes for a robot model to reach its destinations when employing three distinct SLAM algorithms: GMapping, Cartographer SLAM, and SLAM_toolbox. Furthermore, this study addresses indoor human detection and tracking assignments, in which we evaluate the effectiveness of YOLOv5, YOLOv6, YOLOv7, and YOLOv8 models in conjunction with various optimization algorithms, including SGD, AdamW, and AMSGrad. The study concludes that YOLOv8 with SGD optimization yields the most favorable outcomes for human detection. These proposed systems are rigorously validated through experimentation, utilizing a simulated Gazebo environment within the Robot Operating System 2 (ROS2).

Author 1: Hoang Tran Ngoc

Author 2: Nghi Nguyen Vinh

Author 3: Nguyen Trung Nguyen

Author 4: Luyl-Da Quach

Keywords: Indoor robotic; SLAM; ROS2; Robot model; Human detection; YOLO

Paper 31: Optimizing Network Intrusion Detection with a Hybrid Adaptive Neuro Fuzzy Inference System and AVO-based Predictive Analysis

Abstract: Protecting data and computer systems, as well as preserving the accessibility, integrity, and confidentiality of vital information in the face of constantly changing cyberthreats, requires the vital responsibility of detecting network intrusions. Existing intrusion detection models have limits in properly capturing and interpreting complex patterns in network behavior, which frequently leads to difficulties in robust feature selection and a lack of overall intrusion detection accuracy. The drawbacks of current methods are addressed by a unique approach to network intrusion detection presented in this paper. This framework discusses the difficulties presented by changing cyberthreats and the critical requirement for efficient intrusion detection in a society growing more networked by the day. Using a Hybrid Adaptive Neuro Fuzzy Inference System and African Vulture Optimization model with Min-Max normalization and data cleaning on the NSL-KDD dataset, the methodology outlined here overcomes issues with complex network behavior patterns and improves feature selection for precise identification of potential security threats. This approach meets the need for an effective intrusion detection system. Python software is used to implement the suggested model since it is flexible and reliable. The results show a notable improvement in accuracy, with the Hybrid Adaptive Neuro Fuzzy Inference System and African Vulture Optimization model surpassing previous approaches significantly and obtaining an exceptional accuracy rate of 99.3%. The accuracy of the proposed model was improved by African Vulture Optimization, rising from 99.2% to 99.3%. When compared to Artificial Neural Network (78.51%), Random Forest (92.21%), and Linear Support Vector Machine (97.4%), this amazing improvement is clear. When compared to other techniques, the suggested model exhibits an average accuracy gain of about 20.79%.

Author 1: Sweety Bakyarani. E

Author 2: Anil Pawar

Author 3: Sridevi Gadde

Author 4: Eswar Patnala

Author 5: P. Naresh

Author 6: Yousef A. Baker El-Ebiary

Keywords: Network intrusion; cyberthreats; normalization; African vulture optimization; data cleaning

Paper 32: Enhancing Style Transfer with GANs: Perceptual Loss and Semantic Segmentation

Abstract: The goal of artistic style translation is to combine an image's substance with an equivalent image's spirit of innovation. Current approaches are unable to consistently capture complex stylistic elements and maintain uniform stylization over semantic segments, which results in artefacts. Also suggest a novel approach which blends subjective loss algorithms using deep networks of neurons with segmentation using semantics to address these issues. By guaranteeing contextually-aware design distribution together with information preservation, the combination improves general aesthetic correctness during the styling transmission process. With this technique, perceptive components are extracted using both the subject matter and the style photos using previously trained deep neural systems. These components combine to provide perceptive loss coefficients, which are subsequently included into the design of a Generative Adversarial Network (GAN). For offering the representation a better grasp of the meaning contained in any given image, an automatic segmenting module is subsequently implemented. This historical data directs the style transferring process, producing an additional precise and sophisticated transition. The outcomes of our experiments confirm the efficacy of this method and demonstrate improved visual accuracy over earlier approaches. The use of semantic segmentation and loss of perceptual information algorithms together provide a significant 95.6% improvement in visual accuracy. This method effectively overcomes the drawbacks of earlier approaches, providing precise and trustworthy transference of style and constituting a noteworthy advancement in the field of imaginative style transfer. The final output graphics further demonstrate the importance of the recommended approach by deftly integrating decorative elements into functionally significant places.

Author 1: A Satchidanandam

Author 2: R. Mohammed Saleh Al Ansari

Author 3: A L Sreenivasulu

Author 4: Vuda Sreenivasa Rao

Author 5: Sanjiv Rao Godla

Author 6: Chamandeep Kaur

Keywords: Artistic style transfer; Generative Adversarial Networks (GANs); semantic segmentation; visual fidelity; deep Convolutional Neural Networks (deep-CNN)

Paper 33: Securing Patient Medical Records with Blockchain Technology in Cloud-based Healthcare Systems

Abstract: Blockchain technology presents a promising solution to myriad challenges pervasive in the healthcare domain, particularly concerning the secure and efficient management of burgeoning health information technology (HIT) data. This paper delineates a novel blockchain-based approach to enhance various aspects of healthcare management, including data accuracy, drug prescriptions, pregnancy data, supply chain management, electronic health record (EHR) management, and risk data management, with a special emphasis on ensuring secure access, immutable record-keeping, and robust data sharing. We propose a solution focusing on leveraging blockchain technology, particularly utilizing a Hyperledger network within Amazon Web Services (AWS), to securely manage patients' medical records in the cloud. The implemented framework, housed within a Virtual Private Cloud (Amazon VPC) to ensure restricted access and cost-effective resource utilization, underscores advancements in data availability, security, traceability, and sharing, addressing key challenges within healthcare data management, and presenting a scalable, efficient, and secure approach to EHR management in contemporary healthcare contexts.

Author 1: Mohammed K Elghoul

Author 2: Sayed F. Bahgat

Author 3: Ashraf S. Hussein

Author 4: Safwat H. Hamad

Keywords: Security; blockchain; cloud; hyperledger

Paper 34: Hyperchaotic Image Encryption System Based on Deep Learning LSTM

Abstract: This paper introduces an advanced method for enhancing the security of image transmission. It presents a novel color image encryption algorithm that combines hyperchaotic dynamics and deep learning medium and long short-term memory (LSTM) networks. Firstly, the chaotic sequence is generated using the Lorenz hyperchaotic system, then the Lorenz chaotic system is discretized and iteratively processed using the fourth-order Runge-Kutta (RK4) method, and then the deep learning LSTM model is used to transform the chaotic sequence processed by the Lorenz hyperchaotic system into a new sequence for training. Finally, according to the new chaotic signal, the Arnold disruption and Deoxyribo Nucleic Acid (DNA) encoding double disruption diffusion are performed to derive the ultimate encrypted image. Through the analysis of multiple color image simulation experiments, the algorithm presented in this paper can well realize the encryption on color images and can achieve lossless encryption, with strong resistance to differential attack, statistical attack and violent attack. Compared with the literature analysis, the correlation coefficient, information entropy and pixel change rate of this paper are closer to the ideal value, and it has higher security and better encryption effect.

Author 1: Shuangyuan Li

Author 2: Mengfan Li

Author 3: Qichang Li

Author 4: Yanchang Lv

Keywords: Image encryption; Lorenz Chaotic System; LSTM model; deep learning; DNA encoding

Paper 35: Thai Finger-Spelling using Vision Transformer

Abstract: In this paper, we present a finger-spelling recognition system that is based on Thai Sign Language (TFS) and employs a deep learning model called vision transformer. We extracted the 15 characters of the Thai alphabet from publicly available and our collected datasets to establish the recognition system. To train the learning model, we employed four EVA-02 vision transformer models, each of which showed impressive performance across different model sizes. We conducted four experiments to determine the most effective performance model. In Experiment 1, we directly trained the model to compare its performance. In Experiment 2, we used augmentation techniques to generate additional datasets. Experiment 3 utilized the Test-Time Augmentation (TTA) technique to generate test images with random variations. Lastly, in Experiment 4, we used Pseudo-Labelling (labeling labeled and unlabeled data) in each batch to train the model network. Furthermore, we developed a mobile application that collects user image data and provides helpful information related to finger-spelling, such as meanings, gestures, and usage examples.

Author 1: Kullawat Chaowanawatee

Author 2: Kittasil Silanon

Author 3: Thitinan Kliangsuwan

Keywords: Thai finger-spelling; vision transformer; deep learning; image recognition

Paper 36: Investigation of Deep Learning Based Semantic Segmentation Models for Autonomous Vehicles

Abstract: Semantic segmentation plays a pivotal role in enhancing the perception capabilities of autonomous vehicles and self-driving cars, enabling them to comprehend and navigate complex real-world environments. Numerous techniques have been developed to achieve semantic segmentation. Still, the paper emphasizes the effectiveness of deep learning approaches because they have demonstrated impressive capabilities in capturing intricate patterns and features from images, resulting in highly accurate segmentation results. Although various studies have been conducted in literature, there is needed for a careful investigation and analysis of the existing methods, especially in terms of two critical aspects: accuracy and inference time. To address this need for analysis and investigation, the research focuses on three widely-used deep learning architectures: ResNet, VGG, and MobileNet. By thoroughly evaluating these models based on accuracy and inference time, the study aims to identify the models that strike the best balance between precision and speed. The findings of this study highlight the most accurate and efficient models for semantic segmentation, aiding the development of reliable self-driving technology.

Author 1: Xiaoyan Wang

Author 2: Huizong Li

Keywords: Semantic segmentation; autonomous vehicles; deep learning approaches; performance analysis; accuracy; inference time

Paper 37: Smart Cities, Smarter Roads: A Review of Leveraging Cutting-Edge Technologies for Intelligent Event Detection from Social Media

Abstract: The rapidly evolving landscape of smart cities and intelligent transportation systems makes the timely detection of traffic events a critical element for optimizing urban mobility. Furthermore, social media emerges as a valuable source of real-time information, with users acting as active sensors who spontaneously share observations and experiences related to traffic incidents. This review paper offers a comprehensive understanding of the state-of-the-art in traffic event detection from social media. The paper explores leveraging cutting-edge technologies including machine learning, and deep learning with big data technologies and high-performance computing. The discussion unfolds with an in-depth examination of the recent approaches for event detection followed by an exploration of the techniques of spatio-temporal information extraction and sentiment analysis, which are both considered fundamental aspects in enhancing the contextual understanding of traffic events. Further, the review explores the pivotal role of big data technologies in addressing scalability challenges inherent in the vast expanse of social data. The examination encompasses how big data frameworks facilitate efficient storage, processing, and analysis of large-scale social media datasets, thereby empowering machine learning and deep learning models for robust and real-time traffic event detection. Subsequently, the challenges and future directions have been highlighted. Addressing these challenges and leveraging advanced technologies, facilitates the proactive detection and management of these events, paving the way for smart mobility systems.

Author 1: Ebtesam Ahmad Alomari

Author 2: Rashid Mehmood

Keywords: Mobility; smart cities; event detection; social media; big data analytics

Paper 38: A Proposed Roadmap for Optimizing Predictive Maintenance of Industrial Equipment

Abstract: Now-a-days, the maintenance management of industrial equipment, particularly in the aeronautical industry, has evolved into a substantial challenge and a critical concern for the sector. Aeronautical wiring companies are currently grappling with escalating difficulties in equipment maintenance. This paper proposes an intelligent system for the automated detection of machine failures. It assesses predictive maintenance approaches and underscores the significance of sensor selection to optimize outcomes. The integration of Machine Learning techniques with the Industrial Internet of Things (IIoT) and intelligent sensors is presented, showcasing the heightened accuracy and effectiveness of predictive maintenance, especially in the aeronautical industry. The research aims to leverage Predictive Maintenance for enhancing the performance of production machines, predicting their failures, recognizing faults, and determining maintenance dates through the analysis and processing of collected data. Employing sophisticated code, the study emphasizes real-time data collection, data traceability, and enhanced precision in predicting potential failures using Machine Learning. The findings underscore the collaboration between sensors and the synergy of Machine Learning with IIoT, ultimately aiming for sustained reliability and efficiency of predictive maintenance in aeronautical wiring companies.

Author 1: Maria Eddarhri

Author 2: Mustapha Hain

Author 3: Jihad Adib

Author 4: Abdelaziz Marzak

Keywords: Predictive maintenance; intelligent system; aeronautical wiring companies; machine learning; IIoT

Paper 39: Research on 3D Target Detection Algorithm Based on PointFusion Algorithm Improvement

Abstract: With the continuous development of automatic driving technology, the requirements for the accuracy of 3D target detection in complex traffic scenes are getting higher and higher. To solve the problems of low recognition rate, long detection time, and poor robustness of traditional detection methods, this paper proposes a new method based on PointFusion model improvement. The method utilizes the PointFusion network architecture to input 3D point cloud data and RGB image data into the PointNet++ and ResNeXt neural network structures, respectively, and adopts a dense fusion method to predict the spatial offsets of each input point to each vertex in the 3D selection box point by point, to output the 3D prediction box of the target. Experimental results on the KITTI dataset show that compared with the PointFusion network model, the improved PointFusion-based model proposed in this paper improves the 3D target detection accuracy in three different difficulty modes (easy, medium, and hard) and performs best in the medium difficulty mode. These findings highlight the potential of the method proposed in this paper to be applied in the field of autonomous driving, providing a reliable basis for navigating self-driving cars in complex environments.

Author 1: Jun Wang

Author 2: Shuai Jiang

Author 3: Linglang Zeng

Author 4: Ruiran Zhang

Keywords: Neural network; target detection; autonomous driving; PointFusion; deep learning

Paper 40: Beyond the Norm: A Modified VGG-16 Model for COVID-19 Detection

Abstract: The outbreak of Coronavirus Disease 2019 (COVID-19) in the initial days of December 2019 has severely harmed human health and the world's overall condition. There are currently five million instances that have been confirmed, and the unique virus is continuing spreading quickly throughout the entire world. The manual Reverse Transcription-Polymerase Chain Reaction (RT-PCR) test is time-consuming and difficult, and many hospitals throughout the world do not yet have an adequate number of testing kits. Designing an automated and early diagnosis system that can deliver quick decisions and significantly lower diagnosis error is therefore crucial. Recent advances in emerging Deep Learning (DL) algorithms and emerging Artificial Intelligence (AI) approaches have made the chest X-ray images a viable option for early COVID-19 screening. For visual image analysis, CNNs are the most often utilized class of deep learning neural networks. At the core of CNN is a multi-layered neural network that offers solutions, particularly for the analysis, classification, and recognition of videos and images. This paper proposes a modified VGG-16 model for detection of COVID-19 infection from chest X-ray images. The analysis has been made among the model by considering some important parameters such as accuracy, precision and recall. The model has been validated on publicly available chest X-ray images. The best performance is obtained by the proposed model with an accuracy of 97.94%.

Author 1: Shimja M

Author 2: K. Kartheeban

Keywords: Covid-19; coronavirus; artificial intelligence; deep learning; transfer learning; VGG-16; performance metrics

Paper 41: Optimizing Crack Detection: The Integration of Coarse and Fine Networks in Image Segmentation

Abstract: In recent years, the automation of detecting structural deformities, particularly cracks, has become vital across a wide range of applications, spanning from infrastructure maintenance to quality assurance. While numerous methods, ranging from traditional image processing to advanced deep learning architectures, have been introduced for crack segmentation, reliable and precise segmentation remains challenging, especially when dealing with complex or low-resolution images. This paper introduces a novel method that adopts a dual-network model to optimize crack segmentation through a coarse-to-fine strategy. This model integrates both a coarse network, focusing on the global context of the entire image to identify probable crack areas, and a fine network that zooms in on these identified regions, processing them at higher resolutions to ensure detailed crack segmentation results. The foundation of this architecture lies in utilizing shared encoders throughout the networks, which highlights the extraction of uniform features, paired with the introduction of separate decoders for different segmentation levels. The efficiency of the proposed model is evaluated through experiments on two public datasets, highlighting its capability to deliver superior results in crack detection and segmentation.

Author 1: Hoanh Nguyen

Author 2: Tuan Anh Nguyen

Keywords: Deep learning; crack segmentation; coarse-to-fine strategy; image segmentation

Paper 42: Enhanced Land Use and Land Cover Classification Through Human Group-based Particle Swarm Optimization-Ant Colony Optimization Integration with Convolutional Neural Network

Abstract: Reliable classification of Land Use and Land Cover (LULC) using satellite images is essential for disaster management, environmental monitoring, and urban planning. This paper introduces a unique method that combines a Convolutional Neural Network (CNN) with Human Group-based Particle Swarm Optimization (HPSO) and Ant Colony Optimization (ACO) algorithms to improve the accuracy of LULC classification. The suggested hybrid HPSO-ACO-CNN architecture effectively solves the issues with feature selection, parameter optimization, and model training that are present in conventional LULC classification techniques. During the initial phases, HPSO and ACO are crucial in identifying the best hyperparameters for the CNN model and fine-tuning the selection of critical spectral bands. ACO modifies the CNN's hyperparameters (learning rate, batch size, and convolutional layers), whereas HPSO finds the optimal selection of spectral bands. This optimization technique reduces the probability of overfitting while substantially enhancing the model's ability to generalize. Utilizing the selected spectral bands and optimum parameter configuration, the CNN algorithm is trained in the second phase. With Python implementation, this method uses both the spatial and spectral characteristics that the CNN detects to reach an outstanding 99.3% accuracy in LULC classification. The hybrid approach outperforms traditional methods like Deep Neural Network (DNN), Multiclass Support Vector Machine (MSVM), and Long Short-Term Memory (LSTM) in experiments using benchmark satellite image datasets, demonstrating a significant 10.5% increase in accuracy. This hybrid HPSO-ACO-CNN architecture transforms accurate and dependable LULC classification, offering an advantageous instrument for remote sensing applications. It enhances the area of satellite imagery evaluation by combining the advantages of deep learning techniques with optimization algorithms, enabling more accurate mapping of land use and cover for sustainable land management and environmental preservation.

Author 1: Moresh Mukhedkar

Author 2: Chamandeep Kaur

Author 3: Divvela Srinivasa Rao

Author 4: Shweta Bandhekar

Author 5: Mohammed Saleh Al Ansari

Author 6: Maganti Syamala

Author 7: Yousef A.Baker El-Ebiary

Keywords: Land use and land cover; human group-based particle swarm optimization; ant colony optimization; convolutional neural network; satellite image

Paper 43: IAM-TSP: Iterative Approximate Methods for Solving the Travelling Salesman Problem

Abstract: TSP is a well-known combinatorial optimization problem with several practical applications. It is an NP-hard problem, which means that the optimal solution for huge numbers of examples is computationally impractical. As a result, researchers have focused their efforts on devising efficient algorithms for obtaining approximate solutions to the TSP. This paper proposes Iterative Approximate Methods for Solving TSP (IAM-TSP), as a new method that provides an approximate solution to TSP in polynomial time. This proposed method begins by adding four extreme cities to the route, a loop, and then adds each city to the route using a greedy technique that evaluates the cost of adding each city to different positions along the route. This method determines the best position to add the city and the also the best city to be added. The resultant route is further improved by employing local constant permutations. When compared to existing state-of-the-art methods, our experimental results show that the proposed method is more capable of producing high-quality solutions. The proposed approach, with an average approximation of 1.09, can be recommended for practical usage in its current form or as a pre-processing step for another optimizer.

Author 1: Esra’a Alkafaween

Author 2: Samir Elmougy

Author 3: Ehab Essa

Author 4: Sami Mnasri

Author 5: Ahmad S. Tarawneh

Author 6: Ahmad Hassanat

Keywords: Greedy algorithms; TSP; NP-Hard problems; polynomial time algorithms; combinatorial problems; optimization methods

Paper 44: An Overview of Different Deep Learning Techniques Used in Road Accident Detection

Abstract: Every year, numerous lives are tragically lost because of traffic accidents. While many factors may lead to these accidents, one of the most serious issues is the emergency services' delayed response. Often, valuable time is lost due to a lack of information or difficulty determining the location and severity of an accident. To solve this issue, extensive research has been conducted on the creation of effective traffic accident detection and information communication systems. These systems use new technology, such as deep learning algorithms, to spot accidents quickly and correctly and communicate important information to emergency workers. This study provides an overview of current research in this field and identifies similarities among various systems. Based on the review findings, it was found that researchers utilised various techniques, including MLP (Multilayer Perceptron), CNN (Convolutional Neural Network), and models such as DenseNet, Inception V3, LSTM (Long short-term memory), YOLO (You Only Look Once), and RNN (Recurrent Neural Network), among others. Among these models, the MLP model demonstrated high accuracy. However, the Inception V3 model outperformed the others in terms of prediction time, making it particularly well-suited for real-time deployment at the edge and providing end-to-end functionality. The insights gained from this review will help enhance systems for detecting traffic accidents, which will lead to safer roads and fewer casualties. Future research must address several challenges, despite the promising results showcased by the proposed systems. These challenges include low visibility during nighttime conditions, occlusions that hinder accurate detection, variations in traffic patterns, and the absence of comprehensive annotated datasets.

Author 1: Vinu Sherimon

Author 2: Sherimon P. C

Author 3: Alaa Ismaeel

Author 4: Alex Babu

Author 5: Sajina Rose Wilson

Author 6: Sarin Abraham

Author 7: Johnsymol Joy

Keywords: Deep learning; road traffic; road accident detection; MLP; CNN; LSTM; DenseNet; RNN; inception V3

Paper 45: IoT-based Autonomous Search and Rescue Drone for Precision Firefighting and Disaster Management

Abstract: Disaster management is a line of work that deals with the lives of people, such work requires utmost precision, accuracy, and tough decision-making under critical situations. Our research aims to utilize Internet of Things (IoT)-based autonomous drones to provide detailed situational awareness and assessment of these dangerous areas to rescue personnel, firefighters, and police officers. The research involves the integration of four systems with our drone, each capable of tackling situations the drone can be in. As the recognition of civilians and protecting them is a key aspect of disaster management, our first system (i.e., Enhanced Human Identification System) to detect trapped victims and provide rescue personnel the identity of the human located. Moreover, it also leverages an Enhanced Deep Super-Resolution Network (EDSR) x4-based Upscaling technology to improve the image of human located. The second system is the Fire Extinguishing System which is equipped with an inbuilt fire extinguisher and a webcam to detect and put off fire at disaster sites to ensure the safety of both trapped civilians and rescue personnels. The third system (i.e., Active Obstacle Avoidance system) ensures the safety of the drone as well as any civilians the drone encounters by detecting any obstacle surrounding its pre-defined path and preventing the drone from any collision with an obstacle. The final system (i.e., Air Quality and Temperature Monitoring system) provides situational awareness to the rescue personnel. To accurately analyze the area and its safety levels, inform the rescue force on whether to take precautions such as wearing a fire proximity suit in case of high temperature or trying a different approach to manage the disaster. With these integrated systems, Autonomous surveillance drones with such capabilities will improve the equation of autonomous Search and Rescue (SAR) operations to a great extent as every aspect of our approach considers both the rescuer and victims in a region of disaster.

Author 1: Shubeeksh Kumaran

Author 2: V Aditya Raj

Author 3: Sangeetha J

Author 4: V R Monish Raman

Keywords: Search and rescue; firefighting; internet of things; disaster management

Paper 46: Contactless Palm Vein Recognition System with Integrated Learning Approach System

Abstract: Palm Vein Recognition (PVR) is a new biometric authentication technology that provides both security and convenience. This paper describes a contactless PVR system (CPVR) that uses an integrated learning approach (ILA) to recognise the palm veins from the given input images while ensuring user comfort and ease of use. Contactless palm vein scanning technology is used in the proposed system, eliminating the need for physical contact with the scanning device. The proposed method combines advanced feature extraction techniques with a light gradient boosting machine (LightGBM) and transfer learning. A pre-trained model, EfficientNetB1, is used to train the model to extract significant factors from the input PVR images. The proposed method improves user comfort and reduces the risk of cross-contamination in environments where hygiene is critical, such as hospitals, banking, and other secured places. The cutting-edge contactless palm vein scanner captures the unique vein patterns beneath the user's palm without requiring direct physical contact. The proposed ILA illuminates and captures vein patterns using near-infrared (NIR) light, ensuring high accuracy and robustness. The system employs advanced pre-processing techniques and enhanced image segmentation techniques to continuously improve recognition accuracy. It adjusts to changes in the user's vein patterns over time, considering factors like ageing and injuries. The ILA improves the system's ability to adjust palm positioning and lighting changes. The ILA is also a Contactless Palm Vein Recognition System with numerous applications, such as access control, secure authentication for financial transactions, healthcare record access, and more. The system is built to be scalable, allowing organisations to use it in various settings, ranging from small-scale installations to large enterprise-level deployments. Finally, the proposed approach ILA used to recognise accurate users increased the detection rate.

Author 1: Ram Gopal Musunuru

Author 2: T Sivaprakasam

Author 3: G Krishna Kishore

Keywords: Palm Vein Recognition (PVR); Light Gradient Boosting Machine (LightGBM); Transfer Learning; Integrated Learning Approach (ILA)

Paper 47: Linear and Nonlinear Analysis of Photoplethysmogram Signals and Electrodermal Activity to Recognize Three Different Levels of Human Stress

Abstract: All human beings experience different levels of psychological stress during their daily activities, and stress is an integral part of human life. So far, few studies have attempted to identify different levels of stress by analyzing physiological signals. However, it should be noted that developing a practical system for detecting multiple stress levels is a challenging task, and no standard system has been developed for this purpose. Therefore, in the current study, we propose a new detection system based on linear and nonlinear analysis of photoplethysmogram (PPG) and electrodermal activity (EDA) signals to classify three levels of stress (low, medium and high). In the current study, we recorded the physiological signals of EDA and PPG during three trials of a Stroop color word test that induced three levels of stress in 42 healthy male volunteers. Mean, median, standard deviation, variance, skewness, kurtosis, minimum, maximum, and RMS features in the time domain were calculated from physiological signals as linear features. Also, approximate entropy, sample entropy, permutation entropy, Hurst exponent, Katz fractal dimension, Higuchi fractal dimension, Petrosian fractal dimension, detrended fluctuation analysis (DFA), and embedding dimension and time delay parameters from phase space reconstruction of the signals were calculated as nonlinear features. The combination of nonlinear and linear features extracted from both PPG and EDA signals resulted in the highest mean accuracy (88.36%), intraclass correlation (ICC) (98.82%) and F1 (89.24%) values in the classification of three levels of mental stress through multilayer perceptron neural network. Our findings showed that the combination of nonlinear and linear approaches for biological data analysis (PPG and EDA) could help to develop a stress detection system.

Author 1: Yan Su

Author 2: Yuanyuan Li

Author 3: Shumin Zhang

Author 4: Hui Wang

Keywords: Stress detection; biological signal; linear analysis; nonlinear analysis; classification

Paper 48: Development of a Framework for Classification of Impulsive Urban Sounds using BiLSTM Network

Abstract: Urban environments are awash with myriad sounds, among which impulsive noises stand distinct due to their brief and often disruptive nature. As cities evolve and expand, the accurate classification and management of these impulsive sounds become paramount for urban planners, environmental scientists, and public health advocates. This paper introduces a novel framework leveraging the Bidirectional Long Short-Term Memory (BiLSTM) Network for the systematic categorization of impulsive urban sounds. Traditional methodologies often falter in recognizing the nuanced intricacies of such noises. In contrast, the presented BiLSTM-based approach adapts to the temporal variability intrinsic to these sounds, thereby enhancing classification accuracy. The research harnesses an expansive dataset, curated from various urban settings, to train and validate the model. Preliminary findings suggest that our BiLSTM framework outperforms existing models, with a marked increase in both specificity and sensitivity metrics. The outcome of this study holds profound implications for city acoustics management, noise pollution control, and urban health interventions. Moreover, the framework's adaptability paves the way for its application across diverse acoustic landscapes beyond the urban realm. Future endeavors should seek to further optimize the model by integrating more diverse soundscapes and addressing potential biases in data collection.

Author 1: Nazbek Katayev

Author 2: Aigerim Altayeva

Author 3: Bayan Abduraimova

Author 4: Nurgul Kurmanbekkyzy

Author 5: Zhumabay Madibaiuly

Author 6: Bakhytzhan Kulambayev

Keywords: Impulsive sound; machine learning; deep learning; CNN; LSTM; classification

Paper 49: Research on Evaluation Method of Urban Human Settlement Environment Quality Based on Back Propagation Neural Network

Abstract: In order to improve people's living experience, a method for evaluating the quality of urban human settlements based on back propagation neural network is proposed. Firstly, the initial evaluation index system is constructed, the initial evaluation index system is screened, and the final evaluation index system is constructed by using the remaining evaluation indexes. Then, the back propagation neural network is constructed to build an evaluation model, and the evaluation model is trained through the processes of network initialization, hidden layer output calculation and output layer output calculation. Finally, the improved genetic algorithm is used to optimize the back propagation neural network, improve the evaluation performance of the back propagation neural network, and realize the evaluation of human settlements quality. The experimental results show that the accuracy of the evaluation results of urban human settlements quality output by the trained back-propagation neural network model reaches 96.3%, which has a good effect.

Author 1: Siyuan Zhang

Author 2: Wenbo Song

Keywords: Back propagation; neural network; urban human settlements; quality evaluation; morbidity index; genetic algorithm

Paper 50: Basketball Motion Recognition Model Analysis Based on Perspective Invariant Geometric Features in Skeleton Data Extraction

Abstract: The study proposes a recognition method based on skeleton data to address the basketball action recognition, especially those posed by viewpoint changes in videos. The key of this method is to extract geometric features of viewpoint invariance and combine them with spatio-temporal feature fusion techniques. In addition, the study constructs a dynamic topological map of the human skeleton based on long and short-term neural networks to improve the model performance. The experimental results showed that the research method had an average accuracy of 97.85% for Top-5 metrics on the Kinetics dataset and 97.82% for Top-5 metrics on the NTU RGB+D dataset. It is significantly better than the other three state-of-the-art methods. According to the experimental results, it achieves efficient and stable basketball action recognition, which is significantly superior to existing methods. This research not only provides a more efficient method for basketball motion recognition, but also provides valuable references for other sports action recognition fields.

Author 1: Jiaojiao Lu

Keywords: Skeleton data; perspective invariance; geometric features; basketball recognition; spatio-temporal feature fusion

Paper 51: Application of Data Mining Technology with Improved Clustering Algorithm in Library Personalized Book Recommendation System

Abstract: The information construction work of university libraries is becoming increasingly perfect. However, the massive amount of data poses significant challenges to the personalized recommendation of books. Cluster analysis has always been an important research topic in data mining technology, and it has a wide range of application fields. Clustering algorithm is a fundamental operation in big data processing, and it also has good application value in personalized recommendation of library books. To improve the personalized service quality of libraries, this study proposes a clustering algorithm based on density noise application spatial clustering. This study introduced a distance optimization strategy and Warhill algorithm to the proposed algorithm, to improve the difficulties in selecting initial parameter neighborhoods and density thresholds in traditional models, as well as computational complexity. Afterwards, this study will integrate the improved algorithm with the density peak algorithm to further improve the operational efficiency of the model. The performance verification of the model demonstrated superior clustering performance. The average accuracy of the proposed model's recommendation is 98.97%, indicating superiority. The practical application results have confirmed that there is a significant similarity between the books read by the readers and the books read by the target readers, and the effectiveness and feasibility of the proposed model have been verified. Therefore, the proposed model can contribute to the personalized recommendation function of libraries and has certain practical significance.

Author 1: Xiao Lin

Author 2: Wenjuan Guan

Author 3: Ying Zhang

Keywords: Peak density; distance optimization; warhill algorithm; collaborative filtering; book recommendations

Paper 52: Workforce Planning for Cleaning Services Operation using Integer Programming

Abstract: The cleaning services industry in Malaysia faces significant challenges in effectively managing its workforce. Workforce planning, a critical procedure that aligns employee skills with suitable positions at the right time, is becoming increasingly essential across various organizations, including postal delivery and cleaning services. However, the absence of proper workforce planning from management teams has emerged as a primary concern in this sector. This study identifies an opportunity to improve the workforce planning in the cleaning industry by employing an optimization approach that aims to minimize hiring costs. The main objective of this study is to minimize hiring costs in cleaning services operations at a public university in Malaysia. To achieve this, an optimization model based on integer programming was proposed to represent the current situation. Data collection involved interviews and company reports for the purpose of understanding the current conditions comprehensively. Factors influencing hiring costs were meticulously selected, considering the organization's specific situation. Model evaluation was conducted through what-if analysis, which allowed the evaluation of solutions provided by the modified models in three what-if scenarios. The findings indicated that the proposed modified model could assist organizations in improving the workforce planning by optimizing the allocation of resources, reducing hiring costs, and enhancing cleaner performance. This study offers valuable insights for the management of cleaning services, paving the way for more effective and efficient workforce planning practices in the industry.

Author 1: Mandy Lim Man Yee

Author 2: Rosshairy Abd Rahman

Author 3: Nerda Zura Zaibidi

Author 4: Syariza Abdul-Rahman

Author 5: Norhafiza Mohd Noor

Keywords: Workforce planning; cleaning services industry; optimization approach; integer programming

Paper 53: Tailored Expert Finding Systems for Vietnamese SMEs: A Five-step Framework

Abstract: This study addresses the underexplored area of EFSs (EFS) tailored for business applications, with a specific focus on supporting Small and Medium Enterprises (SMEs). The principal objective of this research is to develop an EFS designed to cater to the needs of Vietnamese SMEs. The study methodology involves conducting in-depth interviews with Vietnamese SMEs to ascertain their requirements for Vietnamese EFSs. Subsequently, the research proposes an architectural model for the EFS and proceeds to develop the corresponding system. The EFS operates by collecting and analyzing data from diverse online sources to identify Vietnamese experts and individuals of Vietnamese origin who can provide valuable insights and support to enterprises operating in Vietnam. This research framework is guided by five key Husain (2019)’s issues: 1) Expertise evidence selection, 2) Expert representation, 3) Model building, 4) Model evaluation, and 5) Interaction design. By addressing these issues, the study aims to contribute to the development of an effective EFS tailored to the specific needs of Vietnamese SMEs in their quest to find and engage experts for business growth and innovation.

Author 1: Thi Thu Le

Author 2: Xuan Lam Pham

Author 3: Thanh Huong Nguyen

Keywords: Expert Finding System (EFS); Small and Medium-sized Enterprises (SMEs); experts; Vietnamese expert resources; business expertise identification

Paper 54: A Zero-Trust Model for Intrusion Detection in Drone Networks

Abstract: Today's worldwide introduction of drone fleets in a range of industrial applications has led to numerous network security issues, opening drones up to cyberthreats. In response to these challenges, an innovative approach has been proposed to protect drone fleet networks against potentially dangerous cyberattacks. Indeed, drones are considered as flying computers, and the proposed approach takes into account their complex network structure and communication protocols. The proposed system is designed around a multi-agent architecture, with a hybrid zero-trust detection mechanism against known and emerging cyberthreats. The CICIDS2017 dataset was exploited after performing some essential pre-processing tasks including data cleaning, balancing, binarization and dimension reduction. The proposed approach guaranteed high levels of accuracy and scalability, enabling an effective response to potentially dangerous cyber threat scenarios threatening drone fleets. To evaluate the effectiveness of the proposed system, a test portion of CICIDS2017 was used. The accuracy in recognizing benign network traffic reached 99.99% with a very low false alarm rate, ensuring the system's effectiveness against known and unknown cyber threats. Extensive experimental testing has been carried out on never-before-seen data, highlighting the system's remarkable ability to rapidly recognize cyber threats in real time, thereby enhancing the overall security of drone networks. The contribution of the proposed approach is significant for drone network security, as it introduces a comprehensive model designed to meet the specific security requirements of drone fleets. Finally, the proposed approach offers practical prospects for improving the security of drone applications.

Author 1: Said OUIAZZANE

Author 2: Malika ADDOU

Author 3: Fatimazahra BARRAMOU

Keywords: Fleet of drones; security; zero trust; intrusions; cybersecurity; zero day; Multi-Agent

Paper 55: Flood Prediction using Hydrologic and ML-based Modeling: A Systematic Review

Abstract: Flooding, caused by the overflow of water bodies beyond their natural boundaries, has severe environmental and socioeconomic consequences. To effectively predict and mitigate flood events, accurate and reliable flood modeling techniques are essential. This study provides a comprehensive review of the latest modeling techniques used in flood prediction, classifying them into two main categories: hydrologic models and machine learning models based on artificial intelligence. By objectively assessing the advantages and disadvantages of each model type, we aim to synthesize a systematic analysis of the various flood modeling approaches in the current literature. Additionally, we explore the potential of hybrid strategies that combine both modeling methods' best characteristics to develop more effective flood control measures. Our findings provide valuable insights for researchers and practitioners in the field of flood modeling, and our recommendations can contribute to the development of more efficient and accurate flood prediction systems.

Author 1: A Fares Hamad Aljohani

Author 2: Ahmad. B. Alkhodre

Author 3: Adnan Ahamad Abi Sen

Author 4: Muhammad Sher Ramazan

Author 5: Bandar Alzahrani

Author 6: Muhammad Shoaib Siddiqui

Keywords: Flood prediction; hydrologic model; machine learning; systematic review

Paper 56: An Improved Depth Estimation using Stereo Matching and Disparity Refinement Based on Deep Learning

Abstract: Stereo matching techniques are a vital subject in computer vision. It focuses on finding accurate disparity maps that find its use in several applications namely reconstruction of a 3D scene, navigation of robot, augmented reality. It is a method of obtaining corresponding matching point in stereo images to get disparity map. With additional details, this disparity map could be converted into a depth of a scene. Obtaining an efficient disparity map in the texture less, occluded, and discontinuous areas is a difficult job. A matching cost using an improvised Census transform and an optimization framework is proposed to produce an initial disparity map. The classic Census transform focus on the value of pixel at the center. If this pixel is prone to noisy condition, then the census encoding may differ which leads to mismatches. To overcome this issue an improved census transform based on weighted sum values of the neighborhood pixels is proposed which suppresses the noise during stereo matching. Additionally, a deep learning based disparity refinement technique using the generative adversarial network to handle texture less, occluded, and discontinuous areas is proposed. The suggested method offers cutting-edge performance in terms of both qualitative and quantitative outcomes.

Author 1: Deepa

Author 2: Jyothi K

Author 3: Abhishek A Udupa

Keywords: Census transform; deep learning; depth; generative adversarial network; occlusion; stereo matching

Paper 57: Federated-Learning Topic Modeling Based Text Classification Regarding Hate Speech During COVID-19 Pandemic

Abstract: One of the most challenging tasks in knowledge discovery is extracting the semantics of the content regarding emotional context from the natural language text. The COVID-19 pandemic gave rise to many serious concerns and has led to several controversies including spreading of false news and hate speech. This paper particularly focuses on Islamophobia during the COVID-19. The widespread usage of social media platforms during the pandemic for spreading of false information about Muslims and their common religious practices has further fueled the existing problem of Islamophobia. In this respect, it becomes very important to distinguish between the genuine information and the Islamophobia related false information. Accordingly, the proposed technique in this paper extracts features from the textual content using approaches like Word2Vec and Global Vectors. Next, the text classification is performed using various machine learning and deep learning techniques. The performance comparison of various algorithms has also been reported. After experimental evaluation, it was found that the performance metric like F1-score indicate that Support Vector Machine performs better than other alternatives. Similarly, Convolutional Neural Network also achieved promising results.

Author 1: Muhammad Kamran

Author 2: Ammar Saeed

Author 3: Ahmed Almaghthawi

Keywords: Knowledge extraction; text mining; pandemics and society; hate speech; Islamophobia

Paper 58: A Neural Network-based Approach for Apple Leaf Diseases Detection in Smart Agriculture Application

Abstract: Plant diseases significantly harm agriculture, which has an impact on nations' economies and levels of food security. Early plant disease detection is essential in smart agriculture. For the diagnosis of plant diseases, a number of methods, including imaging, have been used recently. Some of the existing methods for plant disease detection using imaging have limitations as firstly, high computational cost, some methods require complex image processing algorithms or manual design of features that can increase the time and resources needed for the detection. Secondly, low accuracy, most of the methods rely on simple classifiers or handcrafted features that may not capture the subtle differences between different diseases or healthy leaves. Thirdly, dependency on expert knowledge, some methods need human intervention or prior knowledge of the diseases and pests to perform the detection. These limitations are not suitable for the problem at hand because they can affect the efficiency of the detection system. In this study, three apple tree leaf diseases—apple black spot, Alternaria, and Minoz blight—are detected using a neural network (NN) and a digital image processing technique. The sample images are prepared, processed, and used to extract attributes using a digital image processing approach, and the NN is used to classify the diseases. An evaluation of the proposed system's performance in identifying illnesses in apple trees shows satisfactory accuracy and strong overall performance. Additionally, when compared to other techniques already in use, this strategy is more effective at diagnosing.

Author 1: Shengjie Gan

Author 2: Defeng Zhou

Author 3: Yuan Cui

Author 4: Jing Lv

Keywords: Smart agriculture; plant disease; apple leaf disease; image processing; neural network

Paper 59: The Use of Hand Gestures as a Tool for Presentation

Abstract: Our hands play a crucial role in daily activities, serving as a primary tool for interacting with technology. This paper explores using hand gestures to control presentations, offering a dynamic alternative to traditional devices like mice or keyboards. These conventional methods often limit presenters to a fixed position and depend on the device's proximity. In contrast, hand gesture controls promise a more fluid and engaging presentation style. This study utilizes the HaGRID dataset, supplemented by custom-recorded data, divided into 80% for training, and 10% each for validation and testing. The data undergoes preprocessing and a linear classifier with four dense layers and a SoftMax activation layer is employed. The model, optimized with the Adam optimizer and a learning rate of 1e-1, incorporates a motion classifier (LSTM) with two dense layers and an LSTM layer, tailored for long-distance body pose estimation. The resulting application, a local desktop tool independent of internet connectivity, uses tkinter for its user interface. It demonstrates high accuracy in classifying gestures, achieving 90.1%, 89%, and 90% in training, validation, and testing, respectively, for the linear classifier. The motion classifier records 79.8%, 72%, and 70.1%. The model effectively recognizes and categorizes dataset gestures, capturing live camera feeds to manage presentations. Users benefit from various features, including PowerPoint selection, distance mode, gesture toggling and assignment, and appearance mode. This study illustrates how hand gesture control can enhance presentation experiences, merging technology with natural human movement for a more seamless interaction.

Author 1: Hope Orovwode

Author 2: John Amanesi Abubakar

Author 3: Onuora Chidera Gaius

Author 4: Ademola Abdullkareem

Keywords: Hand gesture; linear classifier; motion classifier; LSTM; interface

Paper 60: SmishGuard: Leveraging Machine Learning and Natural Language Processing for Smishing Detection

Abstract: SMS facilitates the transmission of concise text messages between mobile phone users, serving a range of functions in personal and business domains such as appointment confirmation, authentication, alerts, notifications, and banking updates. It plays a vital role in daily communication due to its accessibility, reliability, and compatibility. However, SMS unintentionally generates an environment where smishing can occur. This is because SMS is extensively available and reliable. Smishing attackers exploit this trust to trick victims into divulging sensitive information or performing malicious actions. Early detection saves users from being victimized. Researchers introduced different methods for accurately detecting smishing attacks. Machine Learning models, coupled with Language Processing methods, are promising approaches for combating the escalating menace of SMS phishing attacks by analyzing large datasets of SMS messages to differentiate between legitimate and fraudulent messages. This paper presents two methods (SmishGaurd) to detect smishing attacks that leverage machine learning models and language processing techniques. The results indicate that TF-IDF with the LDA method outperforms Weight Average Word2Vec in precision and F1-Score, and Random Forest and Extreme Gradient Boosting demonstrate higher accuracy.

Author 1: Saleem Raja Abdul Samad

Author 2: Pradeepa Ganesan

Author 3: Justin Rajasekaran

Author 4: Madhubala Radhakrishnan

Author 5: Hariraman Ammaippan

Author 6: Vinodhini Ramamurthy

Keywords: Smishing; phishing; SMS; machine learning; natural language processing; TF-IDF

Paper 61: Sleep Apnea Detection Method Based on Improved Random Forest

Abstract: Random forest (RF) helps to solve problems such as the detection of sleep apnea (SA) by constructing multiple decision trees, but there is no definite rule for the selection of input features in the model. In this paper, we propose a SA detection method based on fuzzy C-mean clustering (FCM) and backward feature rejection method, which improves the sensitivity and accuracy of SA detection by selecting the optimal set of features to input to the random forest model. Firstly, FCM clustering is performed on the RR interval features of ECG signals, and then the backward feature rejection method is used to combine the intra-cluster tightness, inter-cluster separation and contour coefficient metrics to eliminate redundant features to determine the optimal feature set, which is then inputted into the RF to detect SA. The experimental results of this method on Apnea-ECG database data show that the SA detection accuracy is 88.6%, sensitivity is 90.5%, and specificity is 85.5%, and the algorithm can adaptively select a smaller number of more discriminative features through FCM to reduce the input dimensions and improve the accuracy and sensitivity of the RF model for sleep apnea detection.

Author 1: Xiangkui Wan

Author 2: Yang Liu

Author 3: Liuwang Yang

Author 4: Chunyan Zeng

Author 5: Danni Hao

Keywords: Sleep apnea; fuzzy c-means; backward feature elimination method; random forest

Paper 62: Graph Anomaly Detection with Graph Convolutional Networks

Abstract: Anomaly detection in network data is a critical task in various domains, and graph-based approaches, particularly Graph Convolutional Networks (GCNs), have gained significant attention in recent years. This paper provides a comprehensive analysis of anomaly detection techniques, focusing on the importance and challenges of network anomaly detection. It introduces the fundamentals of GCNs, including graph representation, graph convolutional operations, and the graph convolutional layer. The paper explores the applications of GCNs in anomaly detection, discussing the graph convolutional layer, hierarchical representation learning, and the overall process of anomaly detection using GCNs. A thorough review of the literature is presented, with a comparative analysis of GCN-based approaches. The findings highlight the significance of graph-based techniques, deep learning, and various aspects of graph representation in anomaly detection. The paper concludes with a discussion on key insights, challenges, and potential advancements, such as the integration of deep learning models and dynamic graph analysis.

Author 1: Aabid A. Mir

Author 2: Megat F. Zuhairi

Author 3: Shahrulniza Musa

Keywords: Anomaly detection; deep learning; dynamic graphs; Graph Convolutional Networks (GCNs); Graph Neural Networks (GNNs); network data

Paper 63: Ascertaining Speech Emotion using Attention-based Convolutional Neural Network Framework

Abstract: Conversation among people is a profuse form of interaction that also carries emotional information. Speech input has been the subject of numerous studies over the last ten years, and it is now crucial for human-computer connection, as well as for medical care, privacy, and stimulation. This research aims to evaluate if the suggested framework can aid in speech emotion recognition (SER) activities and determine if Convolutional Neural Network (CNN) systems are efficient for SER activities using transfer learning models on spectrogram. In this investigation, the authors present a brand-new attention-based CNN framework and evaluate its efficacy against several well-known CNN architectures from earlier research. The effectiveness of the suggested system is assessed using the SAVEE dataset, an open-access resource for emotive speech, compared to famous CNN models like VGG16, InceptionV3, ResNet50, InceptionResNetV2, and Xception. The authors used stacked 10-fold cross-validation on SAVEE for all of our trials. Amongst these CNN structures, the suggested model had the greatest accuracy (87.14%), followed by VGG16 (83.19%) and InceptionResNetV2 (82.22%). Compared to contemporary techniques, the test results and evaluation show our proposed approach to have steady and impressive results.

Author 1: Ashima Arya

Author 2: Vaishali Arya

Author 3: Neha Kohli

Author 4: Namrata Sukhija

Author 5: Ashraf Osman Ibrahim

Author 6: Salil Bharany

Author 7: Faisal Binzagr

Author 8: Farkhana Binti Muchtar

Author 9: Mohamed Mamoun

Keywords: Convolutional neural network; emotions; speech; transfer learning models; spectrogram

Paper 64: Convolutional LSTM Network for Real-Time Impulsive Sound Detection and Classification in Urban Environments

Abstract: In recent years, the escalating challenges of noise pollution in urban environments have necessitated the development of more sophisticated sound detection and classification systems. This research introduces a novel approach employing a Convolutional Long Short-Term Memory (ConvLSTM) network tailored for real-time impulsive sound detection in metropolitan landscapes. Impulsive sounds, characterized by sudden onsets and short durations—such as honking, abrupt shouts, or breaking glass—are inherently sporadic but can significantly impact urban soundscapes and the well-being of city dwellers. Traditional sound detection mechanisms often falter in identifying these ephemeral noises amidst the cacophony of urban life. The ConvLSTM network proposed in this study amalgamates the spatial feature learning capabilities of Convolutional Neural Networks (CNN) with the temporal sequence retention attributes of LSTM, culminating in an architecture that excels in both sound detection and classification tasks. The model was trained and evaluated on a comprehensive dataset sourced from various urban settings and demonstrated commendable proficiency in discerning impulsive sounds with minimal false positives. Furthermore, the system's real-time processing capabilities ensure timely interventions, paving the way for smarter noise management in cities. This research not only propels the frontier of impulsive sound detection but also underscores the potential of ConvLSTM in addressing multifaceted urban challenges.

Author 1: Aigerim Altayeva

Author 2: Nurzhan Omarov

Author 3: Sarsenkul Tileubay

Author 4: Almash Zhaksylyk

Author 5: Koptleu Bazhikov

Author 6: Dastan Kambarov

Keywords: Deep learning; CNN; LSTM; hybrid model; ANN; impulsive sound

Paper 65: Breast Cancer Detection System using Deep Learning Based on Fusion Features and Statistical Operations

Abstract: Breast cancer is considered as the second cause of death for women. The earlier is diagnosed, the easier the patients can be recovered. The need for studies to detect this kind of cancer easily and accurately came from the growing rate of infected patents by breast cancer exponentially. This study is conducted to investigate the use of deep-learning model for breast cancer detecting using the technique VGG-19 and ultrasound images. Two layers of VGG19 structure were used: (i.e. fc6 and fc7. Based on these two layers (fc6 and fc7), new datasets were created, which are named as statistical operations. These datasets will be employed as input for the following Machine Learning classifiers: K-Nearest Neighbors, Random Forest, Naïve Bayes and Decision Tree. Data augmentation was considered to increase the dataset size for better learning of CNN. Random Forest achieved high accuracy (88.63), precision (0.88), recall (0.88) and F-Measure (0.88). The results of the classification accuracy in the three scenarios are slightly similar; this proves that the breast cancer can be detected even if the size of data in the training dataset was minimal.

Author 1: Suleyman A. AlShowarah

Keywords: Breast cancer detection; breast cancer classification; deep learning; vgg-19; breast tumor

Paper 66: Detecting Threats from Live Videos using Deep Learning Algorithms

Abstract: Threat detection is an important area of research, particularly in security and surveillance applications. The research is focused on developing a threat detection system using DL techniques. The system aims to detect potential threats in real-time video streams, enabling early identification and timely response to potential security risks. The study uses two state-of-the-art DL models, MobileNet and YOLOv5, to train the object detection system. The TensorFlow object detection API is employed for training and evaluating the models. The results of the study indicate that MobileNet outperforms YOLOv5 in terms of detection accuracy, speed, and overall performance. The justification for selecting MobileNet over YOLOv5 is based on several factors. First, MobileNet has a lightweight architecture, making it suitable for real-time applications where processing speed is critical. Second, it is efficient in terms of memory usage, enabling it to operate effectively on low-resource devices. Third, MobileNet provides high accuracy in detecting objects of different sizes and shapes. The study evaluated the performance of the threat detection system using various evaluation metrics, including mean average recall (mAR), mean average precision (mAP) and Intersection over union (IoU). The results show that the system achieved high accuracy in detecting threats, with an overall mAP (mean average precision) of 0.9125, mAR (mean average recall) of 0.9565 and Intersection over union (IoU) of 0.9045. In this study, researchers present a highly efficient and successful method for identifying threats through the utilization of deep learning methods. The research demonstrates the superiority of MobileNet over YOLOv5 in terms of performance, and the results obtained validate the effectiveness of the proposed system in detecting potential threats in real-time video streams.

Author 1: Rawan Aamir Mushabab AlShehri

Author 2: Abdul Khader Jilani Saudagar

Keywords: Deep learning; machine learning; object detection; threat detection

Paper 67: Developing an Improved Method to Remove Pectoral Muscle for Better Diagnosis of Breast Cancer in Mammography Images

Abstract: Mammography is a non-invasive method to study breast tissues for abnormalities. Computer-aided diagnosis (CAD) can automate the process of diagnosing malignant and benign tumors accurately. However, accurate results can be hampered by the presence of the pectoral muscle, which has a similar opacity to the breast tissue area. Detecting and removing pectoral muscles is not trivial due to various factors, and there are artifacts present near the pectoral muscle that can hamper proper segmentation. Given the significance of the topic, it is crucial to devise an accurate method for automatically detecting the muscle area in a mammography image and eliminating it from the rest of the image. This process of removing the pectoral muscle from the breast image can aid in precise segmentation and diagnosis of the tumor area, ultimately leading to faster diagnosis and better outcomes for patients. This study examined two segmentation algorithms, Level Set and Region Growing, for segmenting the pectoral muscle. An Improved Region Growing-based (IRG) algorithm was also proposed and showed promising results in automatically segmenting the pectoral muscle. All algorithms were tested on the MIAS dataset, and radiologists evaluated the results, showing an accuracy rating of up to 83% for IRG. The results indicated that IRG outperformed Level Set considerably due to many optimizations and modifications. IRG can be used as part of the preprocessing unit of an automated cancer diagnosis system.

Author 1: Golnoush Abaei

Author 2: Zahra Rezaei

Author 3: Usama Qasim Mian

Author 4: Yasir Azhari Abdalgadir Abdalla

Author 5: Nitin Mathew

Author 6: Leong Yi Gan

Keywords: Breast cancer; preprocessing pectoral muscle segmentation; level set algorithm; region growing algorithm

Paper 68: Applying Machine Learning Models to Electronic Health Records for Chronic Disease Diagnosis in Kuwait

Abstract: The leading cause of death nowadays is chronic disease. As a result, personal wellbeing has received a considerable boost as a healthcare preventative strategy. A notable development in data-driven healthcare technology is the creation of a prediction model for chronic diseases. In this situation, computational intelligence is used to analyze electronic health data to provide clinicians with knowledge that will help them make more informed decisions about prognoses or therapies. In this study, various classification algorithms have been implemented namely, Decision Tree, K-Nearest Neighbors, Logistic Regression, Multilayer Perceptron, Naïve Bayes, Random Forest, and Support Vector Machines to examine the medical records of patients in Kuwait who had chronic conditions. For predicting diabetes, the support vector machines classifier was the best classifier for predicting diabetes and kidney chronic diseases. For diabetes, it achieved 88.5% accuracy, and 93.6% f1-score, while for kidney; it achieved 94.9% and 92.6% accuracy and f1-score respectively. For predicting heart disease, MLP was the best and achieved 84.7%, and 87.8% accuracy and f1-score respectively.

Author 1: Talal M. Alenezi

Author 2: Taiseer H. Sulaiman

Author 3: Amr M. AbdelAziz

Keywords: Chronic diseases; Electronic Health Records (EHR); machine learning; classification

Paper 69: Separability-based Quadratic Feature Transformation to Improve Classification Performance

Abstract: Feature transformation is an essential part of data preprocessing to improve the predictive performance of machine learning (ML) algorithms. Box-Cox transformation with the goal of separability is proven to align with the performance improvement of ML algorithms. However, the features mapped using Box-Cox transformation preserve the order of the data, so it is ineffective when used to improve the separability of multimodal distributed features. This research aims to build a feature transformation method using quadratic functions to improve class separability that can adaptively change the order of the data when necessary. Fisher score (Fs) measures the separability level by maximizing the Fisher's Criteria of the quadratic function. In addition to increasing the Fs value of each feature, this method can also make the feature more informative, as evidenced by the increasing value of information gain, information gain ratio, Gini decrease, ANOVA, Chi-Square, reliefF, and FCBF. The increase in Fs is particularly significant for bimodally distributed features. Experiments were conducted on 11 public datasets with two statistical-based machine learning algorithms representing linear and nonlinear ML algorithms to validate the success of this method, namely LDA and QDA. The experimental results show an improvement in accuracy in almost all datasets and ML algorithms, where the highest accuracy improvement is 0.268 for LDA and 0.188 for QDA.

Author 1: Usman Sudibyo

Author 2: Supriadi Rustad

Author 3: Pulung Nurtantio Andono

Author 4: Ahmad Zainul Fanani

Keywords: Separability; feature transformation; quadratic function; fisher’s criterion; fisher score

Paper 70: Detecting Data Poisoning Attacks using Federated Learning with Deep Neural Networks: An Empirical Study

Abstract: The advent of intelligent networks powered by machine learning (ML) methods over the past few years has dramatically facilitated various facets of human lives, including healthcare, transportation, and entertainment. However, the use of ML in intelligent networks raises serious concerns about privacy and security, particularly in the context of data poisoning attacks. In order to address these concerns, this research paper presents a novel technique for detecting data poisoning attacks in intelligent networks, focusing on addressing privacy and security concerns associated with the use of machine learning (ML) methods. The research combines federated learning and deep learning approaches to analyze network data in a distributed and privacy-preserving manner. The technique employs a federated neural network to identify malicious data by analyzing network traffic, leveraging the power of Bayesian convolutional neural networks for efficient and accurate detection. The research follows an empirical approach, conducting experimental analyses to evaluate the proposed technique's effectiveness in terms of network security and data classification. The results demonstrate significant performance, including high throughput, quality of service, transmission rate, and low root mean square error for network security. Furthermore, the technique achieves impressive accuracy, recall, precision and malicious data analysis for data detection. The findings of this research contribute to enhancing the security and integrity of intelligent networks, benefiting various stakeholders, including network administrators, data privacy advocates, and users relying on secure network communication.

Author 1: Hatim Alsuwat

Keywords: Poisoning attacks; deep learning; network security; data classification; malicious data

Paper 71: Strengthening AES Security through Key-Dependent ShiftRow and AddRoundKey Transformations Utilizing Permutation

Abstract: AES (Advanced Encryption Standard) is a widely applied block cipher standard in the United States, used in various security applications today. Currently, there are numerous research endeavors aimed at making AES block ciphers dynamic to improve their security against contemporary strong attacks. The most common dynamic approach involves the dynamization of AES block transformations, including SubByte, ShiftRow, AddRoundKey, and MixColumn operations. The combination of these transformations has also been explored and proposed. However, to the best of our knowledge, the dynamic combination of AddRoundKey and ShiftRow transformations remains unexplored. Therefore, in this paper we introduce algorithms for generating key-dependent AddRoundKey and ShiftRow transformations based on permutations. Subsequently, these key-dependent transformations are applied to AES to create dynamic AES block ciphers. Security analysis and evaluation of NIST’s statistical criteria are performed, and the entropy of AES and dynamic AES is assessed. From our findings, it is evident that dynamic AES block ciphers can significantly enhance AES security and meet stringent randomness criteria, similar to AES.

Author 1: Tran Thi Luong

Keywords: AES; ShiftRow; AddRoundKey; dynamic AES; key-dependent

Paper 72: Selection of Unmanned Aircraft Development Model in Indonesia using the AHP Method

Abstract: Countries worldwide are attempting to acquire or create Class 3 unmanned aircraft as part of their armies’ primary weapons systems. The development of medium altitude long endurance (MALE) unmanned aircraft in Indonesia forms part of the national strategic program. Based on documentation studies, three alternative MALE-class unmanned aircraft development models were identified. This study aims to determine the most appropriate unmanned aircraft development model for the MALE class for Indonesia’s current situation. This will aid decision-making by the government and stakeholders related to the drone development model. The analytical hierarchy process (AHP) method was used to analyze the decision-making for the selection of an unmanned aircraft development model. The study began with a questionnaire survey of 11 experts from various institutions. The results show that the priority criterion should be the benefits obtained, followed by the opportunity and budget criteria, and, finally, the risk. The consortium model, which had the highest score of 0.548, is the most suitable for Indonesia’s development of MALE-class unmanned aircraft. The results of the study are expected to provide useful input for AHP researchers, government institutions, and stakeholders.

Author 1: Agus Bayu Utama

Author 2: Siswo Hadi Sumantri

Author 3: Romie Oktovianus Bura

Author 4: Gita Amperiawan

Keywords: Analytical Hierarchy Process (AHP); decision-making; development model; medium altitude long endurance (MALE); unmanned aircraft

Paper 73: Unleashing the Potential of Artificial Bee Colony Optimized RNN-Bi-LSTM for Autism Spectrum Disorder Diagnosis

Abstract: The diagnosis of Autism Spectrum Disorder (ASD) is a crucial, drawn-out, and sometimes subjective procedure that calls for a high level of knowledge. Automation of this diagnostic procedure appears to be possible because to recent developments in machine learning techniques. This paper presents a unique method for improving the performance of a Recurrent Neural Network with a Bidirectional Long Short-Term Memory (RNN-BiLSTM) model for ASD diagnosis by utilizing the power of Artificial Bee Colony (ABC) optimization. Because Python software is used to carry out the implementation, accessibility and adaptability in clinical contexts are guaranteed. The suggested approach is thoroughly contrasted with current techniques, such as ABC optimization for feature extraction, Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) models, and Transfer Learning, in order to highlight its effectiveness. The outcomes demonstrate the superiority of the RNN-BiLSTM over other methods, with much greater accuracy and precision. Combining RNN-BiLSTM with ABC optimization demonstrates not just cutting-edge accuracy but also excellent interpretability. By using this sophisticated model's capabilities, an outstanding diagnosis accuracy of 99.12% is attained, which is 2.77% higher than previous approaches. The model helps physicians comprehend the diagnosis process by highlighting important characteristics and trends that influence its conclusion. Additionally, it lessens the subjectivity and unpredictability involved in human diagnosis, which may result in quicker and more accurate diagnoses of ASD. The research emphasizes how well the Artificial Bee Colony optimized RNN-BiLSTM model diagnoses autism spectrum disorder. By integrating AI-driven diagnostic tools into clinical practice, this research improves early diagnosis and intervention for ASD.

Author 1: Suresh Babu Jugunta

Author 2: Yousef A.Baker El-Ebiary

Author 3: K. Aanandha Saravanan

Author 4: Kanakam Siva Rama Prasad

Author 5: S. Koteswari

Author 6: Venubabu Rachapudi

Author 7: Manikandan Rengarajan

Keywords: Autism spectrum disorder; artificial bee colony; recurrent neural network; bidirectional long short-term network; artificial intelligence

Paper 74: Exploring the Insights of Bat Algorithm-Driven XGB-RNN (BARXG) for Optimal Fetal Health Classification in Pregnancy Monitoring

Abstract: Pregnancy monitoring plays a pivotal role in ensuring the well-being of both the mother and the fetus. Accurate and timely classification of fetal health is essential for early intervention and appropriate medical care. This work presents a novel method for classifying fetal health optimally by combining the Bat Algorithm (BA) in an effective manner with a hybrid model that combines Recurrent Neural Networks (RNN) and Extreme Gradient Boosting (XGB). The Bat Algorithm, inspired by the echolocation behaviour of bats, is employed to optimize the hyperparameters of the XGB-RNN hybrid model. This enables the model to adapt dynamically to the complexities of fetal health data, enhancing its performance and predictive accuracy. The XGB-RNN hybrid model is designed to capitalize on the strengths of both algorithms. XGB provides superior feature selection and gradient boosting capabilities, while RNN excels in capturing temporal dependencies in the data. This approach effectively deals with the difficulties involved in classifying fetal health in the context of pregnancy monitoring by combining these approaches. Python is used to implement the proposed framework. To validate the performance of the proposed approach, extensive experiments were conducted on a comprehensive dataset comprising a wide range of physiological parameters related to fetal health. When it comes to fetal health, BAT Algorithm's XGB-RNN (BARXG) performs outstandingly, greater than other classifiers in terms of accuracy, sensitivity, and specificity. The proposed BARXG model has greater accuracy (98.2%) than existing techniques, which include SVM, Random Forest Classifier, LGBM, Voting Classifier, and EHG.

Author 1: Suresh Babu Jugunta

Author 2: Manikandan Rengarajan

Author 3: Sridevi Gadde

Author 4: Yousef A.Baker El-Ebiary

Author 5: Veera Ankalu. Vuyyuru

Author 6: Namrata Verma

Author 7: Farhat Embarak

Keywords: BAT; fetal health; pregnancy monitoring; RNN; XGBoost

Paper 75: Efficiency Analysis of Firefly Optimization-Enhanced GAN-Driven Convolutional Model for Cost-Effective Melanoma Classification

Abstract: Early identification is essential for successful treatment of melanoma, a potentially fatal type of skin cancer. This work takes a fresh approach to addressing the urgent need for an accurate and economical melanoma categorization system. Inaccuracy, efficiency, and resource usage are common problems with current techniques. A model that incorporates a number of innovative methods to get beyond these restrictions was used in this study. To improve data quality, first applied the pre-processing with a Gaussian filter and augment our dataset with Generative Adversarial Networks (GAN). To extract and classify features, this suggested model makes use of Convolutional Long Short-Term Memory (LSTM) networks. The model performs better and is substantially more accurate when Firefly Optimization is used. It analyses the model's ability to lower healthcare costs by doing a cost-effective analysis, especially when detecting melanoma, including situations involving bleeding lesions. The proposed FFO Enhanced Conv-LSTM's cost-effective analysis makes it possible to compare it favourably to deep convolutional neural networks (DCNN), showcasing its promise for melanoma classification accuracy and healthcare resource allocation optimization. For this study, Python software was used as the implementation tool. The suggested model achieves a 99.1% accuracy rate, which is better than current techniques. A comparative study with well-known models such as Res Net 50, Mobile Net, and Dense Net 169 highlights the notable enhancement provided by the proposed Firefly Optimization-enhanced Conv-LSTM method. This model offers a promising advancement in the precise and economical classification of melanoma due to its high accuracy and cost-effectiveness. In comparison to existing approaches like Res Net 50, Mobile Net, and Dense Net 169, the suggested Firefly Optimization-enhanced Convolutional LSTM (FFO Enhanced Conv-LSTM) method shows an average gain of roughly 5.6% in accuracy.

Author 1: Lakshmi K

Author 2: Sridevi Gadde

Author 3: Murali Krishna Puttagunta

Author 4: G. Dhanalakshmi

Author 5: Yousef A. Baker El-Ebiary

Keywords: Melanoma; cost effective analysis; long short-term memory; firefly optimization; generative adversarial network

Paper 76: Utilizing Multimodal Medical Data and a Hybrid Optimization Model to Improve Diabetes Prediction

Abstract: Diabetes is a major health issue that affects people all over the world. Accurate early diagnosis is essential to enabling adequate therapy and prevention actions. Through the use of electronic health records and recent advancements in data analytics, there is growing interest in merging multimodal medical data to increase the precision of diabetes prediction. In order to improve the accuracy of diabetes prediction, this study presents a novel hybrid optimisation strategy that seamlessly combines machine learning techniques. In order to merge many models in a way that maximises efficiency while enhancing prediction accuracy, the study employs a collaborative learning technique. This study makes use of two separate diabetes database datasets from Pima Indians. A feature selection process is used to streamline error-free classification. A third method known as Binary Grey Wolf-based Crow Search Optimisation (BGW-CSO), which was produced by merging the Binary Grey Wolf Optimisation Algorithm (BGWO) and Crow Search Optimisation (CSO), is provided to further enhance feature selection capabilities. This hybrid optimisation approach successfully solves the high-dimensional feature space challenges and enhances the generalisation capabilities of the system. The Support Vector Machine (SVM) method is used to analyse the selected characteristics. The performance of conventional SVMs is enhanced by the newly created BGW-CSO technique, which optimises the number of hidden neurons within the SVM. The proposed method is implemented using Python software. The suggested BGW-CSO-SVM approach outperforms the current methods, such as Soft Voting Classifier, Random Forest, DMP_MI, and Bootstrap Aggregation, with a remarkable accuracy of 96.62%. Comparing the suggested BGW-CSO-SVM approach to the other methods, accuracy shows an average improvement of around 16%. Comparative evaluations demonstrate the suggested approach's improved performance and demonstrate its potential for real-world use in healthcare settings.

Author 1: A. Leela Sravanthi

Author 2: Sameh Al-Ashmawy

Author 3: Chamandeep Kaur

Author 4: Mohammed Saleh Al Ansari

Author 5: K. Aanandha Saravanan

Author 6: Veera Ankalu. Vuyyuru

Keywords: Diabetes prediction; multimodal medical data; binary grey wolf optimization; crow search optimization; support vector machine

Paper 77: A Hybrid Movies Recommendation System Based on Demographics and Facial Expression Analysis using Machine Learning

Abstract: Cinemas and digital platforms offer an extensive array of content requiring tailored filtering to cater to individual preferences. While recommender systems prove invaluable for this purpose, conventional movie recommendations tend to emphasize specific attributes, leading to a reduction in overall accuracy and reliability. Notably, the extraction process of facial temporal attributes exhibits a suboptimal level of accuracy, thereby influencing the classification of attributes and the overall accuracy of the recommendation system. This article introduces a hybrid recommender system that seamlessly integrates collaborative filtering and content-based methodologies. The system takes into account crucial factors such as age, gender, emotion, and genre attributes. Films undergo an initial categorization based on genre, with a subsequent selection of the most representative genres to ascertain group preferences. Ratings for these selected movies are then predicted and organized in descending order. Employing Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) models, the system achieves real-time extraction of facial attributes, particularly enhancing the accuracy of emotion attribute extraction through sequential processing. The CNN model demonstrates a commendable 55.3% accuracy score, the LSTM model excels with a 59.1% score, while the combined CNN and LSTM models showcase an impressive 60.2% accuracy. The performance of the recommendation system is rigorously evaluated using standard metrics, including precision, recall, and F1-measure. Results underscore the superior performance of the proposed system across various testing scenarios compared to the established benchmark. Nevertheless, it is noteworthy that the precision of the benchmark marginally surpasses the proposed system in the age groups of 8-14 and 15-24.

Author 1: Mohammed Balfaqih

Keywords: Recommender system; movies recommendation; emotion prediction; k-means clustering; deep learning

Paper 78: Analysis of Ransomware Impact on Android Systems using Machine Learning Techniques

Abstract: Ransomware is a significant threat to Android systems. Traditional methods of detection and prediction have been used, but with the advancement of technology and artificial intelligence, new and innovative techniques have been developed. Machine learning (ML) algorithms are a branch of artificial intelligence that have several important advantages, including phishing detection, malware detection, and spam filtering. ML algorithms can also be used to detect ransomware by learning the patterns and behaviors associated with ransomware attacks. ML algorithms can be used to develop detection systems that are more effective than traditional signature-based methods. The selection of the dataset is a crucial step in developing an ML-based ransomware detection system. The dataset should be large, diverse, and representative of the real-world threats that the system will face. It should also include a variety of features that are informative for ransomware detection. This research presents a survey of ML algorithms for ransomware detection and prediction. The authors discuss the advantages of ML-based ransomware detection systems over traditional signature-based methods. They also discuss the importance of selecting a large, diverse, and representative dataset for training ML algorithms. Two datasets are applied during the conducted experiments, which are SEL and ransomware datasets. The experiments are repeated with different splitting ratios to identify the overall performance of each ML algorithm. The results of the paper are also compared to recent methods of ransomware detection and showed high performance of the proposed model.

Author 1: Anfal Sayer M. Al-Ruwili

Author 2: Ayman Mohamed Mostafa

Keywords: Ransomware; machine learning; malware detection; phishing detection; spam filtering

Paper 79: Self-Organizing Control Systems for Nonlinear Spacecraft in the Class of Structurally Stable Mappings

Abstract: In recent developments within the domain of aerospace engineering, there is a burgeoning interest in the autonomous control of nonlinear spacecraft using advanced methodologies. The present research delves deep into the realm of self-organizing control systems tailored for such nonlinear spacecraft, emphasizing its application within the framework of structurally stable mappings. By harnessing the inherent characteristics of structurally stable mappings — often renowned for their resilience to minor perturbations and local modifications — this research endeavors to design a control mechanism that mitigates the challenges presented by the intrinsic nonlinearity of spacecraft dynamics. Initial findings suggest a commendable enhancement in spacecraft maneuverability and robustness against unforeseen disturbances. Furthermore, the employment of self-organization principles leads to an adaptive and resilient system that can reconfigure its control strategies in real-time, basing decisions on immediate environmental feedback. This adaptability, in essence, mimics biological systems that evolve and adapt in the face of challenges. Such a breakthrough in nonlinear spacecraft control not only widens the horizons for space exploration by making missions safer and more efficient but also contributes foundational knowledge to the broader field of nonlinear dynamic system controls. Researchers and practitioners are encouraged to explore this synergistic combination of self-organization and structurally stable mappings to further harness its potential in diverse arenas beyond aerospace.

Author 1: Orisbay Abdiramanov

Author 2: Daniyar Taiman

Author 3: Mamyrbek Beisenbi

Author 4: Mira Rakhimzhanova

Author 5: Islam Omirzak

Keywords: Impulsive sound; machine learning; deep learning; CNN; LSTM; classification

Paper 80: Offensive Language Detection on Online Social Networks using Hybrid Deep Learning Architecture

Abstract: In the digital era, online social networks (OSNs) have revolutionized communication, creating spaces for vibrant public discourse. However, these platforms also harbor offensive language that can proliferates hate speech, cyberbullying, and discrimination, significantly undermining the quality of online interactions and posing severe social implications. This research paper introduces a sophisticated approach to offensive language detection on OSNs, employing a novel Hybrid Deep Learning Architecture (HDLA). The urgency of addressing offensive content is juxtaposed with the challenges inherent in accurately identifying nuanced communications, thus necessitating an advanced model that transcends the limitations of traditional natural language processing techniques. The proposed HDLA model synergistically integrates Convolutional Neural Networks (CNNs) with Long Short-Term Memory (LSTM) networks, capitalizing on the strengths of both methodologies. While the CNN component excels in the hierarchical extraction of spatial features within text data, identifying offensive patterns often concealed in the structural nuances, the LSTM network, adept in processing sequential data, captures the contextual dependencies in user posts over time. This duality ensures a comprehensive analysis of complex linguistic constructs, enhancing the detection accuracy for both overt and covert offensive content. Our research meticulously evaluates the HDLA model using extensive, multi-source datasets reflective of diverse OSN environments, establishing benchmarks against prevailing deep learning models. Results indicate a substantial improvement in precision, recall, and F1-score, demonstrating the model's efficacy in identifying offensive language amidst varying degrees of subtlety and complexity. Furthermore, the model maintains high interpretability, providing insights into the intricate mechanisms of offensive content propagation. Our findings underscore the potential of HDLA in fostering healthier online communities by efficiently curating digital content, thereby upholding the integrity of digital communication spaces.

Author 1: Gulnur Kazbekova

Author 2: Zhuldyz Ismagulova

Author 3: Zhanar Kemelbekova

Author 4: Sarsenkul Tileubay

Author 5: Boranbek Baimurzayev

Author 6: Aizhan Bazarbayeva

Keywords: Offensive language; machine learning; deep learning; social media; detection; classification

Paper 81: Automated Detection of Driver and Passenger Without Seat Belt using YOLOv8

Abstract: The issue of traffic accident fatalities is a serious concern on a global scale, and one of the contributing factors is the failure of drivers to adhere to seat belt usage. A notable challenge arises from the limited availability of law enforcement personnel monitoring this particular issue. In this context, there is a compelling need to implement an automated detection system. The development of this system using YOLOv5 has been done. However, there are weaknesses related to the length of training and detection time. Therefore, this paper proposed a new system using the YOLOv8 method to detect drivers and passengers who violate seat belt regulations. The proposed system is divided into three subsystems: windshield detection, passenger classification, and seat belt classification. YOLOv8 is the latest version of the YOLO (You Only Look Once) method and has been proven to provide better performance than previous versions. Furthermore, this paper also compared five YOLOv8 models, namely YOLOv8n, YOLOv8s, YOLOv8m, YOLOv8l, and YOLOv8x. The proposed model is trained and tested using image data collected from several roads in Indonesia. The experiment results show that the YOLOv8s model produced the best mean Average Precision (mAP) of 0.960 for windshield detection. YOLOv8s-cls and YOLOv8l-cls models achieved the same accuracy of 0.8923 for passenger classification. The YOLOv8l-cls model produced the best accuracy of 0.8846 for seat belt classification. In addition, the proposed method can increase mAP and training time for windshield detection compared to YOLOv5.

Author 1: Sutikno

Author 2: Aris Sugiharto

Author 3: Retno Kusumaningrum

Keywords: Windshield detection; passenger classification; seat belt classification; YOLOv8

Paper 82: Enhancing Alzheimer's Disease Diagnosis: The Efficacy of the YOLO Algorithm Model

Abstract: The diagnosis and early detection of Alzheimer's Disease (AD) and other forms of dementia have become increasingly crucial as our aging population grows. In recent years, deep learning, particularly the You Only Look Once (YOLO) architecture, has emerged as a promising tool in the field of neuroimaging and machine learning for AD diagnosis. This comprehensive review investigates the recent advances in the application of YOLO for AD diagnosis and classification. We scrutinized five research papers that have explored the potential of YOLO, delving into the methodologies, datasets, and results presented. Our review reveals the remarkable strides made in AD diagnosis using YOLO, while also highlighting challenges, such as data scarcity and research lacking. The paper provides insights into the growing role of YOLO in the early detection of AD and its potential to transform clinical practices in the field. This review aims to inspire further research and innovation to enhance AD diagnosis and, ultimately, patient care.

Author 1: Tran Quang Vinh

Author 2: Haewon Byeon

Keywords: Machine learning; deep learning; YOLO; alzheimer’s disease; dementia

Paper 83: Enhancing IoT Security and Privacy with Claims-based Identity Management

Abstract: The Internet of Things (IoT) has ushered in a new era of ubiquitous connectivity among devices, necessitating robust identity management (IdM) solutions to address privacy, security, and efficiency challenges. In this study, it delve into various IdM approaches in the context of IoT, examining their implications for privacy preservation, user experience, integration, and efficiency. In this paper a methodology is an innovative holistic IdM system that leverages emerging cryptographic technologies and a claims-based approach. This system empowers both users and smart objects to manage data disclosure via partial identities and efficient proof mechanisms, ensuring privacy while facilitating seamless interactions which integrate the proposed IdM system with Distributed Capability-Based Access Control (DCapBAC) and Ciphertext-Policy Attribute-Based Encryption (CP-ABE) to cater to diverse IoT scenarios. Through a comparative evaluation, it is highlighted that the limitations of conventional IdM methods and OAuth-based approaches, underscored by the superior efficiency exhibited by our proposed system. Notably efficient, the IdM system stands as a paramount solution for ensuring secure, private, and resource-effective interactions within the ever-expanding IoT landscape. As the IoT domain continues to evolve, embracing advanced identity management systems like our proposal becomes indispensable for fostering trust, bolstering security, and optimizing interactions across interconnected devices and services.

Author 1: Mopuru Bhargavi

Author 2: Yellamma Pachipala

Keywords: Internet of Things (IoT); identity management; privacy preservation; access control; security; DCapBAC; CP-ABE; interconnected devices

Paper 84: Speech Enhancement using Fully Convolutional UNET and Gated Convolutional Neural Network

Abstract: Speech Enhancement aims to enhance audio intelligibility by reducing background noises that often degrade the quality and intelligibility of speech. This paper brings forward a deep learning approach for suppressing the background noise from the speaker's voice. Noise is a complex nonlinear function, so classical techniques such as Spectral Subtraction and Wiener filter approaches are not the best for non-stationary noise removal. The audio signal was processed in the raw audio waveform to incorporate an end-to-end speech enhancement approach. The proposed model's architecture is a 1-D Fully Convolutional Encoder-to-Decoder Gated Convolutional Neural Network (CNN). The model takes the simulated noisy signal and generates its clean representation. The proposed model is optimized on spectral and time domains. To minimize the error among time and spectral magnitudes, L1 loss is used. The model is generative, denoising English language speakers, and capable of denoising Urdu language speech when provided. In contrast, the model is trained exclusively on the English language. Experimental results show that it can generate a clean representation of a clean signal directly from a noisy signal when trained on samples of the Valentini dataset. On objective measures such as PESQ (Perceptual Evaluation of Speech Quality) and STOI (Short-Time Objective Intelligibility), the performance evaluation of the research outcome has been conducted. This system can be used with recorded videos and as a preprocessor for voice assistants like Alexa, and Siri, sending clear and clean instructions to the device.

Author 1: Danish Baloch

Author 2: Sidrah Abdullah

Author 3: Asma Qaiser

Author 4: Saad Ahmed

Author 5: Faiza Nasim

Author 6: Mehreen Kanwal

Keywords: Speech enhancement; speech denoising; deep neural network; raw waveform; fully convolutional neural network; gated linear unit

Paper 85: Hotspot Identification Through Pick-Up and Drop-Off Analysis of Ride-Hailing Transport Service

Abstract: It is important to extract hotspots in urban traffic networks to improve driver route efficiency. This research aims to identify hotspot pick-up and drop-off (PUDO) areas in ride-hailing transportation services using a clustering approach. However, there are challenges in applying clustering algorithms to trajectory data in the coordinates of the Global Positioning System (GPS). So this research proposes modifications to the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm by considering the radius from the center of the cluster to determine the presence of amenities around the cluster. We used a dataset containing 55,988 trip trajectories of Grab drivers over a two-week period in Jakarta. A preliminary statistical analysis was carried out to understand the distribution of trips. Next, we identify the PUDO point of each trip for use in the clustering analysis. The research explores the various parameters and settings of the clustering method and their impact on the results. The study found that the results obtained from the clustering method are sensitive to parameter selection, including epsilon radius and minimum number of points needed to form a cluster. The optimal cluster with the best parameters (eps: 0.25, minpts: 100) in the pick-up (PU) location analysis produced 17 clusters with the silhouette coefficient of 0.752, while in the drop-off (DO) location there are 18 clusters with a silhouette coefficient of 0.694. Overall, the research highlights the potential of the clustering analysis method for ride-hailing transportation.

Author 1: Ragil Saputra

Author 2: Suprapto

Author 3: Agus Sihabudin

Keywords: Hotspot identification; ride-hailing; transportation; PUDO location; clustering analysis

Paper 86: Learning Engagement of Children with Dyslexia Through Tangible User Interface: An Experiment

Abstract: This paper presents the evaluation of a mobile application employing Tangible User Interface (TUI) technology to enhance the educational involvement of children experiencing dyslexia. The primary objective of this application is to assist these children in overcoming challenges related to reading, spelling, pronunciation, and writing, issues often associated with lower self-esteem and dissatisfaction in an academic setting. The study adopts a User-Centered Design (UCD) approach, focusing on the specific needs and preferences of children with dyslexia during development. The evaluation involved 30 children with dyslexia, divided into two groups: a control group utilizing the non-tangible DisleksiaBelajar mobile app (DB) and a treatment group utilizing the DisleksiaBelajar 3D Tangible (DB3dT) app, which incorporates tangible elements. Results indicated that the DB3dT app achieved significantly higher usability scores (79.5%) compared to the DisleksiaBelajar app (51%). Furthermore, the treatment group utilizing the DB3dT app surpassed the control group in learning performance. In summary, the evaluation demonstrated that integrating tangible elements into the DB3dT app notably enhanced the learning experience for children with dyslexia when compared to the non-tangible DisleksiaBelajar app. The children exhibited increased engagement and a willingness to repeat activities, suggesting potential advancements in learning outcomes and performance.

Author 1: Siti Nurliana Jamali

Author 2: Novia Admodisastro

Author 3: Azrina Kamaruddin

Author 4: Saadah Hassan

Keywords: Dyslexia; Tangible User Interface; mobile application; user centered design; engagement

Paper 87: FOREX Prices Prediction Using Deep Neural Network and FNF

Abstract: One of the largest financial markets on the planet is the foreign exchange (FOREX) market. Banks, retail traders, businesses, and individuals trade more than $5.1 trillion in FOREX daily. It is very challenging to predict prices in advance due to the market's complex, volatile, and highly fluctuating nature. In this study, the new FOREX Normalization Function (FNF) is proposed and used with different models to predict the prices of the AUD/USD, EUR/USD, USD/JPY, CHF/INR, USD/CHF, AUD/JPY, USD/CAD, and GBP/USD. Two models are proposed in this study. The first model contains FNF as a normalization and feature extractor, followed by a Convolutional Neural Network (CNN). The second model utilizes FNF and a Support Vector Regressor(SVR). The forecasts are set for a one-day timeframe, with predictions made for 1, 3, 7, and 15 days ahead. The efficient ability of the proposed method to solve the FOREX prediction problem is proven by performing experiments on nine real-world datasets from different currencies. Additionally, the models are evaluated using Mean Absolute Error (MAE) and Mean Squared Error (MSE). Applying the presented models to 9 different datasets improved the results by an average between 0.5% and 58% of MAE.

Author 1: Asmaa M. Moustafa

Author 2: Mohamed Waleed Fakhr

Author 3: Fahima A. Maghraby

Keywords: FOREX prediction; CNN; normalization function; SVR

Paper 88: A New Steganography Method for Hiding Text into RGB Image

Abstract: Now-a-days, the network has significant roles in transferring data and knowledge quickly and accurately from sender to receiver. However, the data is still not secure enough to transfer quite confidentially. Data protection is considered as one of the principal challenges in information sharing over communication. So, steganography techniques were proposed which are the art of hiding information that prevents secret text message detection from intruders. Nevertheless, most steganography methods use low bits number of secret messages. Moreover, these methods applied a single logic gate for encrypting the secret message. Therefore, this paper proposes a new method for the encryption of secret messages based on the Huffman technique to reduce the secret message dimensions. In addition, the proposed method uses two different logic gates namely XOR and XNOR for increasing the message security. The RGB Lena image is used as the cover image of the secret message. There are six different experiments conducted with respect to various lengths of the secret messages in bits. The experimental results show that when using the highest number of bits (i.e., 66288), the proposed method achieved 0.0233 MSE, 64.4589 PSNR, 0.9999998 SSIM, and 8.2383 encryption time. The proposed method has the ability to encrypt the secret message with a high number of bits.

Author 1: AL-Hasan Amer Ibrahim

Author 2: Ruaa Shallal Abbas Anooz

Author 3: Mohammed Ghassan Abdulkareem

Author 4: Musatafa Abbas Abbood Albadr

Author 5: Fahad Taha AL-Dhief

Author 6: Yaqdhan Mahmood Hussein

Author 7: Hatem Oday Hanoosh

Author 8: Mohammed Hasan Mutar

Keywords: Steganography techniques; color images; XOR gate; NOR gate; huffman technique

Paper 89: Bidirectional Long Short-Term Memory for Analysis of Public Opinion Sentiment on Government Policy During the COVID-19 Pandemic

Abstract: One of the initiatives adopted by the Indonesian government to combat the development of COVID-19 in Indonesia is Community Activities Restrictions Enforcement. Many public opinions emerged, both for and against this policy. There are so many comments every second that it is certainly not easy to analyze them by reading each one by one. This task necessitates computer applications. Therefore, this study was conducted to produce an application that can help analyze public sentiment on the policy through social media, namely Twitter, into three classes: positive, neutral, and negative. The method used in this research is bidirectional long short-term memory (BiLSTM), one of the algorithms of deep learning. This study trains the model using the dataset, which consists of 10,486 tweets. The model receives an f1-score of 76.67 %. Thus, the model can be used to analyze public sentiment when the same policy is enforced. It can determine public acceptance of this policy. Thus, the system created in this research can be used as evaluation material for the government to review the policy when it is implemented in the future. However, this study concentrates on how to develop the sentiment analysis system and does not examine how the community responds to government policy.

Author 1: Intan Nurma Yulita

Author 2: Ahmad Faaiz Al-Auza’i

Author 3: Anton Satria Prabuwono

Author 4: Asep Sholahuddin

Author 5: Firman Ardiansyah

Author 6: Indra Sarathan

Author 7: Yusa Djuyandi

Keywords: Sentiment analysis; COVID-19; BiLSTM; deep learning; government policy

Paper 90: New AHP Improvement using COMET Method Characteristic to Eliminate Rank Reversal Phenomenon

Abstract: Rank Reversal in Multi-Criteria Decision Making (MCDM) is a phenomenon that occurs when an alternative is added or deleted because of a change in the order in which the result is ranked. The evaluation of the weight of criteria, which are established based on whether a decision maker considers them important, impacts the alternative ranking result in MCDM. Changes in decision result ranking called rank reversal cannot be acceptable. Many researchers have done lots of research and created new methods for eliminating rank reversal, but until now there is still research that denies these new methods are free from rank reversal. The Analytical Hierarchy Process Method (AHP), the oldest Decision support Method has an advantage in the decision according to the Decision Maker’s (DM’s) preference. Still, it is vulnerable to the rank reversal phenomenon. While Characteristic Object Method (COMET) is a method claimed to be free of rank reversal phenomenon. This paper will discuss how the integration of COMET to AHP especially in the phase of generating characteristic value and characteristic objects is added to the AHP phase, which will have an impact on digital marketing strategy decision-making for private Universities in Indonesia, especially the city of Palembang. The combination of COMET and AHP in this paper is tested with several testing tools; they are case study testing, accuracy testing, and sensitivity analysis testing. The result of the combination of COMET and AHP will be named C-AHP, which is a consideration of DM’s preference for the criteria weight, and the generation of alternative comparison based on criteria, or any other attributes makes AHP free from rank reversal.

Author 1: Yulistia

Author 2: Ermatita

Author 3: Samsuryadi

Author 4: Abdiansah

Keywords: Method; combination; C-AHP; rank reversal; elimination

Paper 91: Automated Detection and Classification of Soccer Field Objects using YOLOv7 and Computer Vision Techniques

Abstract: In the last two decades, many technologies have been deployed and utilized in Soccer games (Football) as a result to the huge investment of Federation of International Football Association (FIFA). These technologies aim to monitor and track all soccer match objects including players and the ball itself in order to measure the player performance, and tracking the players’ positions and movements at the field. Latest emerging artificial intelligence and computer vision techniques are being used recently in many systems and deployed in different scenarios. Identifying all field objects automatically has to be the first step in the monitoring process of soccer games. In this paper, we are proposing an automated system that has the ability to detect and track the ball and to detect and classify players and referees on the soccer field. The proposed system implements a detection model using a real-time object detection model YOLOv7 to detect the ball and all humans on the field after building a labeled dataset of 1300 different soccer game frames. It also deploys Improved Color Coherence Vector (ICCV) features to classify all humans on the field to five classes (Team1, Team2, Goalkeeper1, Goalkeeper2, and Referee) using K-Nearest Neighbor algorithm. The proposed system has achieved high accuracy in both the detection and classification modules.

Author 1: Jafar AbuKhait

Author 2: Murad Alaqtash

Author 3: Ahmad Aljaafreh

Author 4: Waleed Othman

Keywords: Soccer game; football; YOLOv7; human detection and classification; ball detection; improved color coherence vector

Paper 92: Quality In-Use of Mobile Geographic Information Systems for Data Collection

Abstract: Mobile Geographic Information Systems (GIS) plays a vital role in data collection, offering diverse functionalities for spatial data handling. Despite advancements, accurately determining the usage environment during development remains challenging. This study uses machine learning and natural language processing to automatically classify user reviews based on the ISO 25010 quality-in-use model. Motivated by the challenge of gauging user experience during development, stakeholders analyze user reviews for insights. An experimental study compares Support Vector Machine (SVM), Random Forest, Logistic Regression, and Naive Bayes classifiers, revealing superior performance by SVM and Random Forest, particularly in efficiency evaluation. Findings underscore the efficacy of SVM in classifying user reviews, emphasizing its effectiveness in evaluating efficiency within mobile GIS applications. Moreover, it provides valuable insights for stakeholders, contributing to the enhancement of software quality of mobile GIS apps.

Author 1: Badr El Fhel

Author 2: Ali Idri

Keywords: Mobile GIS for data collection; machine learning; software product quality; ISO/IEC 25010; natural language processing; user experience

Paper 93: Bitcoin Optimized Signal Allocation Strategies using Decomposition

Abstract: Bitcoin is the first and most famous cryptocurrency. It is a virtual currency that is operated in a decentralized form using cryptographic strategies called blockchains. Although it has experienced significant market acceptance by traders and investors in recent years, it also suffers from volatility and riskiness. Technical analysis is one of the most powerful tools used for trading signals’ allocation using some algorithmic strategies called technical indicators. In this research, a newly proposed multi-objectives decomposition-based particle swarm optimization algorithm is used to find the best parameter values for some technical indicators, which in turn generates the best trading signals for Bitcoin trading. In this context, three conflicting objectives have been used, i.e., the return on investment, the Sortino-ratio, and the number of trades. The proposed algorithm is compared to the original MOEA/D algorithm as well as the indicators using their original parameters. Results showed the superiority of the proposed algorithm during the training and testing periods over the other benchmarks.

Author 1: Sherin M. Omran

Author 2: Wessam H. El-Behaidy

Author 3: Aliaa A. A. Youssif

Keywords: Bitcoin; technical analysis; decomposition; particle swarm optimization; MOEA/D

Paper 94: A New Method for Revealing Traffic Patterns in Video Surveillance using a Topic Model

Abstract: Research on video surveillance systems, for instance, in intelligent transportation systems, has advanced due to the growing requirement for monitoring, control, and intelligent management. One of the next issues is extracting patterns and automatically classifying them, given the volume of data produced by these systems. In this study, a theme approach was utilized to translate visual patterns into visual words in order to reveal and extract traffic patterns at crossings. The supplied video is first cut up into segments. The optical flux technique is then used to determine the clips' optical flux characteristics, which are based on a lot of local motion vector data, and translate them into visual words. The thin-group thematic coding method is then used to teach traffic patterns to the proposed system using a non-probable thematic model. By responding to a behavioral query like "Where is a vehicle going?" these patterns convey observable motion that can be utilized to characterize a scene. The results of applying the suggested method to the QM_UL video database demonstrated that the suggested method can accurately identify and depict significant traffic patterns such as left turns, right turns, and intersection crossings.

Author 1: Yao Wang

Keywords: Group thin topic coding; QM_UL video; optical flux; traffic patterns

Paper 95: Improving Deep Reinforcement Learning Training Convergence using Fuzzy Logic for Autonomous Mobile Robot Navigation

Abstract: Autonomous robotic navigation has become hotspot research, particularly in complex environments, where inefficient exploration can lead to inefficient navigation. Previous approaches often had a wide range of assumptions and prior knowledge. Adaptations of machine learning (ML) approaches, especially deep learning, play a vital role in the applications of navigation, detection, and prediction about robotic analysis. Further development is needed due to the fast growth of urban megacities. The main problem of training convergence time in deep reinforcement learning (DRL) for mobile robot navigation refers to the amount of time it takes for the agent to learn an optimal policy through trial and error and is caused by the need to collect a large amount of data and computational demands of training deep neural networks. Meanwhile, the assumption of reward in DRL for navigation is problematic as it can be difficult or impossible to define a clear reward function in real-world scenarios, making it challenging to train the agent to navigate effectively. This paper proposes a neuro-symbolic approach that combine the strengths of deep reinforcement learning and fuzzy logic to address the challenges of deep reinforcement learning for mobile robot navigation in terms of training time and the assumption of reward by incorporating symbolic representations to guide the learning process, and inferring the underlying objectives of the task which is expected to reduce the training convergence time.

Author 1: Abdurrahman bin Kamarulariffin

Author 2: Azhar bin Mohd Ibrahim

Author 3: Alala Bahamid

Keywords: Autonomous navigation; deep reinforcement learning; mobile robots; neuro-symbolic; Fuzzy Logic

Paper 96: Brain Tumor Segmentation Algorithm Based on Asymmetric Encoder and Multimodal Cross-Collaboration

Abstract: To address the challenges of insufficient multimodal information fusion and insufficient long-range dependencies features extraction for brain tumor segmentation, this paper propose a novel network based on asymmetric encoder and multimodal cross-collaboration. The network employs an asymmetric encoder-decoder architecture. Firstly, the invert ConvNext split convolution (ICSC) block is used in the local refinement encoder and improved SwinTransformer with DscMLP enhancements (DscSwinTransformer) module is used in global associative encoder. The local and long-range dependencies of each stage of two parallel encoders can be well extracted by hybrid fusion. Moreover, this paper adds a multimodal cross-collaboration (MCC) module at the beginning of the two encoders to fully exploit the complementary information between modalities and reduce the reliance on a single modality during model training. Coordinate Attention (CA) is used in the bridge part of the encoder and decoder to capture important spatial location information. Then, the depthwise separable convolution (DscConv) module is used in the decoder branch to reduce the computation while maintaining good feature extraction ability. Finally, this paper uses a hybrid loss function of BCE, Dice and L2 loss to mitigate the problem of class datas imbalance. Experimental results show that our model achieves Dice coefficients of 0.897, 0.905 and 0.824 in the whole, core and enhanced tumor regions, respectively. These results show that the performance of our proposed method outperforms in comparison with several existing methods in core and enhanced tumor regions.

Author 1: Pengyue Zhang

Author 2: Qiaomei Ma

Keywords: Brain tumor; multimodal cross-collaboration; asymmetric encoder; coordinate attention

Paper 97: Blockchain Integrated Neural Networks: A New Frontier in MRI-based Brain Tumor Detection

Abstract: Brain tumors originating from uncontrolled growth of abnormal cells in the brain, presents a significant challenge in healthcare due to their various symptoms and infrequency. While Magnetic Resonance Imaging (MRI) is essential for accurately identifying and diagnosing malignant tumors, manual interpretation is often complex and sensitive to mistakes. To address this, we introduce BrainTumorNet, a specialized convolutional neural network (CNN) created for MRI-based brain tumor diagnosis. We ensure improved image quality and a robust dataset for model training by including preprocessing approaches involving CLAHE and data augmentation. Additionally, we integrated a blockchain-based data retrieval technology to enhance the security, traceability, and collaboration in MRI data management across several medical institutions. This blockchain framework ensures that MRI data, once input from hospitals, stays immutable and can be safely retrieved based on unique hospital IDs, promoting a trustable environment for data exchange. Performance assessments conducted on multiple MRI datasets showcased BrainTumorNet’s commendable proficiency, with accuracy rates of 98.66%, 97.17% and 94.24% on the dataset 1, dataset 2, and dataset 3, respectively. The model’s performance was evaluated using a comprehensive set of metrics, including accuracy, specificity, recall, precision, f1-score, and confusion matrix. These measures are essential for evaluating a model's strengths and limits, emphasizing BrainTumorNet’s ability to generate accurate and relevant predictions and its effectiveness in determining negative classification. BrainTumorNet's performance was compared with six renowned deep learning architectures: VGG16, ResNet50, AlexNet, MobileNetV2, InceptionV3, and DenseNet121. Our work highlights BrainTumorNet's potential capabilities in simplifying and boosting the accuracy of MRI-based brain tumor diagnosis while ensuring data integrity and collaboration through blockchain.

Author 1: Subrata Banik

Author 2: Nani Gopal Barai

Author 3: F M Javed Mehedi Shamrat

Keywords: Brain tumor; MRI imaging; BrainTumorNet; deep learning; image classification; augmentation

Paper 98: Proposal of a Machine Learning-based Model to Optimize the Detection of Cyber-attacks in the Internet of Things

Abstract: In this article, we propose a model to optimize the detection of attacks in IoT. IoT network is a promising technology that connects living and non-living things around the world. Despite the increased development of these technologies, cyber-attacks remains a weakness, making it vulnerable to numerous cyber-attacks. Of course, automatic computer intrusion detection systems are deployed. However, it does not make it possible to mobilize the full potential of Machine Learning. Our approach in this maneuver consists of offering a means to select the least expensive ML method in terms of learning in order to optimize the prediction of threats to introduce IoT objects. To do this, we make modular design based on two layers. The first module is a canvas containing the different methods most used in ML such as supervised learning method, unsupervised learning method and reinforcement learning method. The second module introduces a mechanism to measure the learning cost linked to each of these methods in order to choose the least expensive one in order to quickly and efficiently detect intrusions in IoT objects. To prove the validity of the proposed model, we simulated it using the Weka tool. The results obtained illustrate the following behaviors: The classification quality rate is 93.66%. This last result is supported by a classification consistency rate of 0.882 (close to unity 1) demonstrating a trend towards convergence between observation and prediction.

Author 1: Cheikhane Seyed

Author 2: Jeanne roux BILONG NGO

Author 3: Mbaye KEBE

Keywords: IoT; Machine learning; cyber-security; detection of attacks; weka tool; classification quality and consistency

Paper 99: Construction of an Intelligent Evaluation Model of Yield Risk Based on Empirical Probability Distribution

Abstract: In order to improve the accuracy of yield risk evaluation, an intelligent evaluation model of yield risk based on empirical probability distribution is constructed. The dimensionality reduction method of risk factor based on principal component analysis is adopted. After adjusting the multiple data dimensions of risk factors that affect the rate of return to a unified dimension, the cluster-based evaluation index screening method is used to build the evaluation index set that best reflects the risk of the rate of return; The index weight vector equation method based on entropy weight and information entropy is used to set the evaluation index weight. Through the comprehensive evaluation model based on the empirical probability distribution of risk indicators, the empirical probability distribution information of risk indicators at all levels is analyzed, and the risk level of yield is intelligently evaluated. The research structure shows that the model can effectively evaluate the level of return risk and provide an effective reference for preventing and controlling investment return risk.

Author 1: Zhou Yanru

Author 2: Yang Jing

Keywords: Empirical probability distribution; yield; risk intelligence evaluation; principal component analysis; clustering; weight

Paper 100: Enhancing Question Pairs Identification with Ensemble Learning: Integrating Machine Learning and Deep Learning Models

Abstract: The effectiveness of machine learning (ML) and deep learning (DL) models on the Quora question pairs dataset is investigated in this study. ML models, including AdaBoost, reached 73.44% test accuracy, while ensemble learning approaches enhanced outcomes even further, with the Hard-Voting Ensemble achieving 76.13%. DL models, such as FCN, demonstrated test accuracy of 81% with cross validation. These findings contribute to natural language processing by demonstrating the potential of ensemble learning for ML models and the DL models' detailed pattern-capturing capacity.

Author 1: Salsabil Tarek

Author 2: Hatem M. Noaman

Author 3: Mohammed Kayed

Keywords: Ensemble learning; natural language processing; deep learning; machine learning

Paper 101: The Fusion Method of Virtual Reality Technology and 3D Movie Animation Design

Abstract: To further improve the design effect of 3D film and television animation, integrating virtual reality technology with 3D film and television animation design is studied. This method uses 3Ds Max software in virtual reality technology to build 3D film and television animation scenes by manual modeling. Based on the established 3D film and television animation scene, texture mapping is performed on it, and then the 3D film and television animation character model is established and simulated. After optimizing the established 3D scene and character model using the improved quadratic error measurement algorithm, the roaming interaction of 3D film and television animation scene is realized through Unity3D software, and the integration of virtual reality technology and 3D film and television animation design is realized. The experimental results indicate that the 3D film and television animation scene created using virtual reality technology is very realistic, which can effectively optimize the 3D film and television animation model. The number of path nodes is the least when the 3D film and television animation scene roams and interacts, which has a relatively significant application effect.

Author 1: Xiang Yuan

Author 2: He Huixuan

Keywords: Virtual reality technology; 3D film and television; Animation design; model optimization; roaming interaction

Paper 102: Classification Method of Traditional Art Painting Style Based on Color Space Transformation

Abstract: In order to improve the accuracy and efficiency of traditional art painting style classification, a classification method of traditional art painting style based on color space transformation is proposed. This method preprocesses the traditional artistic painting style, improves the contrast of the image, makes the color and details of the image more vivid, and provides the basis for the subsequent color space conversion. After the traditional artistic painting style is stretched by automatic contrast stretching method, the color space is transformed. The purpose is to transform the image from one color space to another, so as to better extract the features of the image. Based on the traditional artistic painting style image after color space conversion, the traditional artistic painting image is balanced by the adaptive histogram equalization method with limited contrast, and an enhanced traditional artistic painting image is obtained, which further enhances the contrast of the image, makes the details in the image more prominent, and also enhances the overall visual effect of the image. Taking the enhanced traditional art painting images as input, the fuzzy C-means method is used to classify the traditional art painting styles, and the images are effectively divided into different categories according to the characteristics of the images. The experimental results show that this method can effectively enhance the image of traditional art paintings and effectively classify traditional art paintings with different styles, which has strong application effect.

Author 1: Xu Zhe

Keywords: Color space; traditional art; painting style; classification method; fuzzy c-means

Paper 103: Research on Image Algorithm for Face Recognition Based on Deep Learning

Abstract: As people's requirements for applications are getting higher and higher, the recognition of facial features has been paid more and more attention. The current facial feature recognition algorithm not only takes a long time, but also has problems such as large system resource consumption and long running time in practical applications. Based on this, the research proposes a multi-task face recognition algorithm by combining multi-task deep learning on the basis of convolutional neural network, and analyzes its performance in four dimensions of face identity, age, gender, and fatigue state. The experimental results show that the multi-task face recognition algorithm model obtained through layer-by-layer progression takes less time than other models and can complete more tasks in the same training time. At the same time, comparing the best model M44 with other algorithms in four dimensions, it is found that the Mean Absolute Error lowest is 3.53, and the highest Accuracy value is 98.3%. On the whole, the multi-task face recognition algorithm proposed in the study can recognize facial features efficiently and quickly. At the same time, its training time is short, the calculation speed is fast, and the recognition accuracy is much higher than other algorithms. It is applied to intelligent driving behavior. Analysis, intelligent clothing navigation and other aspects have strong practical significance.

Author 1: Qiang Wu

Keywords: Multi task deep learning; face recognition; convolution neural network; multi task; dimension

Paper 104: A Model for Analyzing Employee Turnover in Enterprises Based on Improved XGBoost Algorithm

Abstract: To accurately predict the possibility of employee turnover during enterprise operation and improve the benefits created by talents in the enterprise, research based on the limit gradient enhancement algorithm has received widespread attention. However, with the exponential growth of various types of resignation reasons, this algorithm is not comprehensive enough when dealing with complex character psychology. To solve this problem, this study uses the limit gradient enhancement algorithm to predict employee turnover in the Company dataset, and uses differential automatic regression moving average variable optimization to generate a fusion algorithm. The research first involves stepwise regression processing of the training data, expanding the objective function to a second-order Taylor expansion; Then variance coding is added to the square integrable linear white noise, and the step cooling curve is smoothed by changing the temperature control constant; Then to calculate the root mean square error of Newton's law of cooling, and obtain its derivative loss variable. Linear white noise is the chaotic data produced by the improved extreme gradient lifting algorithm in forecasting the original data of enterprise employees, which will affect the results of data preprocessing in the loss analysis. In order to reduce the operation error of the algorithm, the step cooling curves are drawn according to the cooling law, and then their root mean square errors are calculated. Finally, the fusion algorithm studied was applied to the Company dataset and the prediction accuracy of the particle swarm optimization algorithm was tested and compared with the fusion algorithm. A total of 400 experiments were conducted, and the fusion algorithm achieved a prediction accuracy of 398 times, with an accuracy rate of 99.5%; The accuracy of particle swarm optimization algorithm is close to that of fusion algorithm, at 83.2%. The experimental results indicate that the algorithm model proposed in the study can accurately predict the possibility of employee turnover in enterprises, and the company will also receive timely information to make the next budget step.

Author 1: Linzhi Nan

Author 2: Han Zhang

Keywords: Data preprocessing; linear white noise; root mean square error; newton’s law of cooling; step cooling curve

Paper 105: Intelligent Design of Ethnic Patterns in Clothing using Improved DCGAN for Real-Time Style Transfer

Abstract: In view of the problems that traditional real-time style transmission technology requires a large number of sample map training, low image quality, lack of realism and detail, this study combines the improved generative adversarial network (GANs) with real-time style transfer technology, and enhances the real-time style transfer calculation with adaptive instance normalization. As a result, a novel intelligent clothing ethnic pattern design model is developed. Experimental results show that the model reduces physical memory usage by 45.7%, with only 453MB, and utilizes only 26% of CPU resources in terms of CPU usage. The training time is approximately 20 minutes and 48 seconds. This model performance is obviously higher than other models. The designed intelligent clothing ethnic pattern design model in this study demonstrates higher clarity and shorter processing time, and has potential applications in the field of image generation.

Author 1: Yingjun Liu

Author 2: Ming Wu

Keywords: Computer vision; improved DCGAN; style transfer; adaptive instance normalization; intelligent design of patterns

Paper 106: AI-Driven Optimization Approach Based on Genetic Algorithm in Mass Customization Supplying and Manufacturing

Abstract: Numerous artificial intelligence (AI) techniques are currently utilized to identify planning solutions for supply chains, which comprise suppliers, manufacturers, wholesalers, and customers. Continuous optimization of these chains is necessary to enhance their performance. Manufacturing is a critical stage within the supply chain that requires continuous optimization. Mass Customization Manufacturing is one such manufacturing type that involves high-volume production with a wide variety of materials. However, genetic algorithms have not been used to minimize both time and cost in the context of mass customization manufacturing. Therefore, we propose this study to present an artificial intelligence solution using genetic algorithm to build a model that minimizes the time and cost which associated with mass customized orders. Our problem formulation is based on a real-world case, and it adheres to expert descriptions. Our proposed optimization model incorporates two strategies to solve the optimization problem. The first strategy employs a single objective function focused on either time or cost, while the second strategy applies the multi-objective function NSGAII to optimize both time and cost simultaneously. The effectiveness of the proposed model was evaluated using a real case study, and the results demonstrated that leveraging genetic algorithms for mass customization optimization outperformed expert estimations in finding efficient solutions. On average, the evaluation revealed a 20.4% improvement for time optimization, a 29.8% improvement for cost optimization, and a 25.5% improvement for combined time and cost optimization compared to traditional expert optimization.

Author 1: Shereen Alfayoumi

Author 2: Neamat Eltazi

Author 3: Amal Elgammal

Keywords: Mass customization manufacturing; metaheuriatic search; genetic algorithm; optimization; supply chain management

Paper 107: Application Model Construction of Emotional Expression and Propagation Path of Deep Learning in National Vocal Music

Abstract: Emotional expression is important in Chinese national vocal music art. The emotional expression in national vocal music is based on the art of national vocal music, with distinct characteristics and requirements. The ultimate goal is to spread the expression of various emotions in the national vocal music art. Promoting the spread of national vocal music singing art using modern media is an urgent requirement for the inheritance and development of national vocal music singing art. With the rapid development of science and technology, integrating deep learning and traditional music has become the general trend. It has been gradually applied to melody recognition, intelligent composition, virtual performance, and other aspects of traditional music and has achieved good results, but also hidden behind a series of ideas and technical and ethical issues. In this paper, the application of deep learning has been discussed and prospected. The recognition rate of emotional expression in national vocal music is 92 %. In terms of communication, combined with the deep learning algorithm, this paper analyzes the characteristics and requirements of emotional expression in the art of national vocal music singing and puts forward a new method of promoting the development of the art of national vocal music singing, hoping to attract more attention and enhance the social awareness of the application field, to promote the steady development of Chinese traditional music in the information age.

Author 1: Zhangcheng Tang

Keywords: Deep learning; national vocal music; innovation; emotion; dissemination

Paper 108: Using Generative Adversarial Networks and Ensemble Learning for Multi-Modal Medical Image Fusion to Improve the Diagnosis of Rare Neurological Disorders

Abstract: The research suggests a unique ensemble learning approach for precise feature extraction and feature fusion from multi-modal medical pictures, which may be applied to the diagnosis of uncommon neurological illnesses. The proposed method makes use of the combined characteristics of Convolutional Neural Networks and Generative Adversarial Networks (CNN-GAN) to improve diagnostic accuracy and enable early identification. In order to do this, a diverse dataset of multi-modal patient medical records with rare neurological disorders was gathered. The multi-modal pictures are successfully combined using a GAN-based image-to-image translation technique to produce fake images that effectively gather crucial clinical data from different paradigms. To extract features from extensive clinical imaging databases, the research employs trained models using transfer learning approaches with CNN frameworks designed specifically for analyzing medical images. By compiling unique traits from each modality, a thorough grasp of the core pathophysiology is produced. By combining the strengths of several CNN algorithms using ensemble learning techniques including voting by majority, weight averaging, and layering, the forecasts were also integrated to arrive at the final diagnosis. In addition, the ensemble approach enhances the robustness and reliability of the assessment algorithm, resulting in increased effectiveness in identifying unusual neurological conditions. The analysis of the collected data shows that the proposed technique outperforms single-modal designs, demonstrating the importance of multi-modal fusion of pictures and feature extraction. The proposed method significantly outperforms existing methods, achieving an accuracy of 99.99%, as opposed to 85.69% for XGBoost and 96.12% for LSTM. The proposed method significantly outperforms existing methods, achieving an average increase in accuracy of approximately 13.3%. The proposed method was implemented using Python software.

Author 1: Bhargavi Peddi Reddy

Author 2: K Rangaswamy

Author 3: Doradla Bharadwaja

Author 4: Mani Mohan Dupaty

Author 5: Partha Sarkar

Author 6: Mohammed Saleh Al Ansari

Keywords: Multi-modal medical images; ensemble learning; CNN; GAN; neurological disorders; image-to-image method; transfer learning; feature extraction

Paper 109: Creating a Framework for Care Needs Hub for Persons with Disabilities and Senior Citizens

Abstract: Patient satisfaction is an assessment that assesses how effectively a company’s goods or services fulfil consumer expectations. This study aims to design an architectural framework for a care needs hub for people with disabilities and senior citizens. Using systems modelling for crafting architectural frameworks, the researchers used a 4+1 view model with UML to intensively describe the features of the care needs hub. Quality attributes were used to indicate how well the system would satisfy the needs of the stakeholders beyond its basic functions. The design includes the system's functional and non-functional features, as well as their corresponding diagrams drawn in a unified modelling language in accordance with the 4+1 view model, to assist the system's developer in mapping the system's functionalities correctly and accurately. Architecture models and design patterns are developed and executed to understand how the system's primary components fit together, how messages and data move effectively across the system, and how other structural issues work. The proposed model includes verified and validated development paradigms and architectural and design patterns that may help accelerate the development process. The architecture and design patterns fulfil all of the system's criteria. The researchers designed a comprehensive tool for the completion of the development of the care needs hub, which would greatly help the developers of the system in crafting the correct features and data abstractions needed to build and implement the said system. This research aims to develop an innovative solution that addresses the current challenges faced by persons with disabilities and senior citizens in accessing care services and provides a comprehensive and accessible platform for their care.

Author 1: Guillermo V. Red

Author 2: Thelma D. Palaoag

Author 3: Vince Angelo E. Naz

Keywords: Care need framework; persons with disability; CareAide; 4+1 view model; CareNeed

Paper 110: Implementation of Cybersecurity Situation Awareness Model in Saudi SMEs

Abstract: Saudi Small and Medium-sized Enterprises (SMEs) are witnessing rapid growth in technology and innovation. However, this growth is accompanied by increased cybersecurity threats, which pose significant challenges for SMEs. Cyber threats are becoming more complex and sophisticated, with SMEs becoming prime targets due to their weaker cybersecurity defenses. Hence, there exists a rich literature on critical challenges facing SMEs. Existing literature on these challenges addresses many research issues (e.g., finance, technology adoption, and management) associated with SMEs. However, one critical issue that has so far received no rigorous attention is cybersecurity situation awareness for research in the SME context. Thus, this study used a quantitative approach aiming to empirically test a model of cybersecurity situational awareness that can support SMEs in Saudi Arabia to implement cybersecurity measures and precautions with efficacy. An online survey of 350 participants was conducted to collect the research data. The study identified a significant positive relationship between Cyber Situational Awareness (Csa) and Implementation of Cybersecurity Controls (Icsc), suggesting that enhancing awareness can contribute to better control implementation. The study identified a significant positive relationship between Cyber Situational Awareness (Csa) and Implementation of Cybersecurity Controls (Icsc), suggesting that enhancing awareness can contribute to better control implementation. Finally, the paper provides several interesting findings and outlines future research directions.

Author 1: Monerah Faisal Almoaigel

Author 2: Ali Abuabid

Keywords: Cyber situation awareness; cybersecurity control and precaution; Saudi; SMEs

Paper 111: Network Security Detection Method Based on Abnormal Traffic Detection

Abstract: To discover potential risks and vulnerabilities in the network in time and ensure the safe operation of the network, a network security detection method based on abnormal traffic detection is studied. Construct network security detection architecture from several aspects, including the front-end interface module, control center module, network status extraction module, anomaly detection module, alarm module, and database module. Use NetFlow technology to capture network traffic from the network in the form of flow, and use the KNN algorithm in the traffic filtering submodule to filter network traffic packets and eliminate duplicate traffic data. After filtering traffic, the traffic data is transmitted to the feature selection sub-module. PCA-TS algorithm is used to reduce the dimension of the network traffic data and select the network traffic characteristics, and then it is input into the SVM classifier. The improved SVM multi-classification algorithm is used to classify normal and abnormal traffic, complete abnormal traffic detection, and achieve network security detection. Experimental results show that the time for feature selection of this method does not exceed 3.0s, and the G score in the detection process also remains above 0.70, indicating that this method has strong network security detection capability.

Author 1: Tao Xiao

Author 2: Yang Ke

Author 3: Hu YiWen

Author 4: Wang HongYa

Keywords: Abnormal traffic; network security detection; data dimensionality reduction; flow characteristics; traffic capture; alarm module

Paper 112: ODFM: Abnormal Traffic Detection Based on Optimization of Data Feature and Mining

Abstract: The booming of computer networks and software applications has led to an explosive growth in the potential damage caused by network attacks. Efficient detection of abnormal traffic in networks is appealing for facilely mastering the traffic tracking and locating for network usage at low resource cost. High quality abnormal traffic detection of Internet becomes particularly relevant during the automated services of multiple application situations. This paper proposes a novel abnormal traffic detection algorithm called ODFM based on the optimization of data feature and mining. Specially, we develop a feature selection strategy to reduce the feature analysis dimension, and set a peer-to-peer (P2P) traffic identification module to filter and mine the related service traffic to reduce the amount of data detection and facilitate the abnormal traffic detection. Experimental results demonstrate that the proposed algorithm greatly improves the detection accuracy, which verifies its effectiveness and competitiveness in the general tasks of abnormal network traffic detection.

Author 1: Xianzong Wu

Keywords: Abnormal traffic; detection; data mining; feature dimension optimization; network security

Paper 113: Automatic Bangla Image Captioning Based on Transformer Model in Deep Learning

Abstract: Indeed, Image Captioning has become a crucial aspect of contemporary artificial intelligence because it has tackled two crucial parts of the AI field: Computer Vision and Natural Language Processing. Currently, Bangla stands as the seventh most widely spoken language globally. Due to this, image captioning has gained recognition for its significant research accomplishments. Many established datasets are found in English but no standard datasets in Bangla. For our research, we have used the BAN-Cap dataset which contains 8091 images with 40455 sentences. Many effective encoder-decoder and Visual Attention approaches are used for image captioning where CNN is utilized for the encoder and RNN is used for the decoder. However, we suggested a transformer-based image captioning model in this study with different pre-train image feature extraction models like Resnet50, InceptionV3, and VGG16 using the BAN-Cap dataset and find out its effective efficiency and accuracy based on many performances measured methods like BLEU, METEOR, ROUGE, CIDEr and also find out the drawbacks of others model.

Author 1: Md. Anwar Hossain

Author 2: Mirza AFM Rashidul Hasan

Author 3: Ebrahim Hossen

Author 4: Md Asraful

Author 5: Md. Omar Faruk

Author 6: AFM Zainul Abadin

Author 7: Md. Suhag Ali

Keywords: Bangla image captioning; image processing; natural language processing; attention mechanism; transformer model

Paper 114: The Hybrid Jaro-Winkler and Manhattan Distance using Dissimilarity Measure for Test Case Prioritization Approach

Abstract: Software product line (SPL) is a concept that has revolutionized the software development industry. It refers to a set of related software products that are developed from a common set of core assets but can be customized to meet specific customer requirements. Integrating SPL techniques into test case prioritization (TCP) can greatly enhance its effectiveness. By considering variability across different products within an SPL, it becomes possible to prioritize test cases based on their relevance to specific product configurations. However, the concept itself still has certain issues, such as in finding the highest rate of early failure detection. Various solutions have been proposed to mitigate this problem, among them is to improve the calculation of string distance using hybrid technique to achieve a high degree for similarity. Dissimilarity-based Technique (DBP) is the basis for our ranking method. The objective is to identify further weaknesses in the product lines as well as the differences between the experiment and real-world applications. Our focus is to enhance hybrid techniques that produce the highest rate of early failure detection. In this paper, early fault detection is selected as the performance goal. In order to choose the optimal methods for DBP for TCP, a comparison between several string distance measures was conducted. This study proposed hybrid techniques that combined Jaro-Winkler and Manhattan string distance namely New Enhanced Hybrid Technique 1 (NEHT1), New Enhanced Hybrid Technique 2 (NEHT2) and New Enhanced Hybrid Technique 3 (NEHT3). The case study was generated using the PLEDGE tool based on a Feature Model (FM). Six test cases were used in the experiment. Result shows the effectiveness of the combination where it achieved higher degree of similarity for T1 vs. T4, T2 vs. T3, T2 vs. T6, and T3 vs. T6, as well as perfect degree of similarity for NEHT1 (100.00%). The result proves that the combination of both techniques improve SPL testing effectiveness compared to existing techniques.

Author 1: Siti Hawa Mohamed Shareef

Author 2: Rabatul Aduni Sulaiman

Author 3: Abd Samad Hasan Basari

Keywords: Test case prioritization; software product line; dissimilarity-based technique; string distance; new enhanced hybrid

Paper 115: A Novel CNN-based Model for Medical Image Registration

Abstract: The registration of the deformable image is applied widely to image diagnosis, the monitoring of the disease, and the navigation of the surgery with the aim of learning the correspondence of the anatomist among an image of motion and an image of static. The procedure of the registration of an image mainly includes three steps: the creation of a model of the deformation, a function design for the mensuration of the similarity, and the step of learning for the optimization of the parameter. In the current article, 2-stream architecture is designed, which has the ability to sequentially estimate the fields of the registration of the multi-level by a couple of the pyramids of the feature. In this paper, a 3D network of the encoder-decoder with the 2-stream is designed, which calculates 2 pyramids of the feature of the convolutional as separately by 2 volumes of the input. Also, the registration of the pyramid of the sequential is proposed, which in it, a trail of the modules of the pyramid registration (PR) for the prediction of the fields of the registration of the multi-level is designed as straight by the pyramids of the feature of the decoding. In addition, the modules of PR can be augmented with the computation of the 3D correlations of the local among the pyramids of the feature, which this work leads to the further improvement of the presented approach. Thus, it is capable of collecting the detailed anatomical structure of the brain. The proposed method is tested in three criterion datasets about the registration of MRI of the brain. The evaluation outcomes display that the presented approach outperforms the advanced approaches with a big value.

Author 1: Hui GAO

Author 2: Mingliang LIANG

Keywords: Image registration; convolutional neural network; Pyramid Registration (PR); encoder-decoder

Paper 116: Recognition of Depression from Video Frames by using Convolutional Neural Networks

Abstract: The disturbances of the mood are relevant to the emotions. Specifically, the behaviour of persons with disturbances of mood, like the depression of the unipolar, displays a powerful correlation of the temporal by the emotional girths of the arousal and the valence. Moreover, the psychiatrists and the psychologists take into account the audible signs of the facial and the audible signs of the voice when they assess the condition of the patient. Depression makes audible behaviours like weak expressions, the validation of the contact of the eye and the use of little flat-voiced sentences. Artificial intelligence has combined various automated frameworks for the detection of depression severity by using hand-crafted features. The method of deep learning has been successfully applied to detect depression. In the current article, a federate architecture, which is the network of the neural of the deep convolutional basis on the attention of global, is proposed to diagnose the depression. This method uses CNN with the attention mechanism and also uses the integration of the weighted spatial pyramid pooling for the learning of the deep global representation. In this method, two branches are introduced: the CNN based on local attention focuses on the patches of the local, while the CNN based on global attention attains the universal patterns from the whole face area. For taking the data of the supplementary among two parts, a CNN basis on the local-global attention is proposed. The designed experiments have been done in two datasets, which are AVEC2014 and AVEC2013. The results show that our presented approach can extract the depression patterns from the video frames. Also, the outcomes display that our presented approach is superior to the best methods based on the video for the detection of depression.

Author 1: Jianwen WANG

Author 2: Xiao SHA

Keywords: Deep learning; depression recognition; Convolutional Neural Network (CNN); attention mechanism

Paper 117: MG-CS: Micro-Genetic and Cuckoo Search Algorithms for Load-Balancing and Power Minimization in Cloud Computing

Abstract: Cloud computing has emerged as a transformative technology, offering remote access to various computing resources. However, efficiently managing these resources while curbing escalating energy consumption remains a critical challenge. In response, this paper presents the Micro-Genetic Algorithm with Cuckoo Search (MG-CS), a novel approach for enhancing cloud computing efficiency. MG-CS optimizes load balancing and power reduction and significantly contributes to reducing operational costs, ensuring compliance with service level agreements, and enhancing overall service quality. Our experiments showcase MG-CS's versatility in achieving a well-balanced distribution of workloads, resource optimization, and substantial energy savings. This multifaceted approach redefines cloud resource management, offering an environmentally sustainable and cost-effective solution. By introducing MG-CS, this research addresses the pressing challenges in cloud computing, aligning it with environmental responsibility and economic efficiency.

Author 1: Jun ZHOU

Author 2: Youyou Li

Keywords: Resource utilization; cloud computing; energy consumption; optimization

Paper 118: A Focal Loss-based Multi-layer Perceptron for Diagnosis of Cardiovascular Risk in Athletes

Abstract: Cardiovascular diseases (CVDs) are a prevalent cause of heart failure around the world. This research was required in order to investigate potential approaches to treating the disease. The article presents a focal loss (FL)-based multi-layer perceptron called MLP-FL-CRD to diagnose cardiovascular risk in athletes. In 2012, 26,002 athletes were measured for their height, weight, age, sex, blood pressure, and pulse rate in a medical exam that had electrocardiography at rest. Outcomes were negative for the largest majority, leading to class imbalance. Training on imbalanced data hurts classifier performance. To address this, the study proposes a training approach based on focal loss, which effectively emphasizes minority class examples. Focal loss softens the influence of simplistic samples, enabling the model to concentrate on more intricate examples. It is useful in circumstances when there is a substantial class imbalance. Additionally, the paper highlights a challenge in the training phase, often characterized by the use of gradient-based learning methods like backpropagation. These methods exhibit several disadvantages, including sensitivity to initialization. The paper recommends the implementation of a mutual learning-based artificial bee colony (ML-ABC). This approach adjusts the primary weight by substituting the food resource candidate, which is selected due to superior fitness, with one based on a mutual learning factor between two individuals. The sample obtains great outcomes, outperforming other machine learning samples. Optimal values for important parameters are identified for the model based on experiments on the study dataset. Ablation studies that exclude FL and ML-ABC of the sample confirm the additive effect of, which is not negative and dependent, these factors on the sample’s efficiency.

Author 1: Chuan Yang

Keywords: Cardiovascular diseases; multi-layer perceptron; focal loss; artificial bee colony; imbalanced classification

Paper 119: Fuzzy Neural Network Algorithm Application in User Behavior Portrait Construction

Abstract: With the increasing number of online users, constructing user behavior profiles has received widespread attention from relevant scholars. In order to construct user behavior profiles more accurately, the research first designed an adaptive fuzzy neural network algorithm based on the momentum gradient descent method. It uses momentum gradient descent to optimize and learn the parameters adjusted by error backpropagation algorithm and least squares estimation method and optimizes the structure of the fuzzy neural network through subtraction clustering. Finally, the improved algorithm is applied to the construction of user behavior profiles. The results showed that in error analysis, the error range of the improved algorithm was within [-0.10, 0.10], and the accuracy was relatively high. In indicator calculation, the improved algorithm had a recall rate of 0.07 and 0.09 higher than the other two algorithms, an accuracy rate of 0.03 and 0.07 higher than the other two algorithms, and an F1 score of 0.07 and 0.08 higher than the other two algorithms, indicating good overall performance. In the ROC curve, the average detection rate of the designed user behavior profiling model was 0.065 and 0.155 higher than the other two models, respectively, with higher detection accuracy. These results demonstrated the effectiveness of improved algorithms and design models, providing certain reference value for the development of related fields.

Author 1: Peisen Song

Author 2: Bengcheng Yu

Author 3: Chen Chen

Keywords: User behavior profiling; momentum gradient descent method; adaptive fuzzy neural network; error backpropagation algorithm; least squares estimation method; subtractive clustering

Paper 120: Automated Classification of Multiclass Brain Tumor MRI Images using Enhanced Deep Learning Technique

Abstract: The brain is a vital organ, and the brain tumor is one of the most dangerous types of tumors in the world. Neuroimaging is an interesting and important discussion in diagnosing central nervous system tumors. Brain tumors have several types, namely meningioma, glioma, pituitary, schwannoma, and neurocytoma. A radiologist uses magnetic resonance imaging (MRI) to detect brain tumors because of its advantages over computed tomography. However, classifying multiclass MRI is difficult and takes a long time. This study proposes an automated classification of multiclass brain tumors using enhanced deep learning techniques. Various models are used in this research, namely VGG16, NasNet-Mobile, InceptionV3, ResNet50, and EfficientNet. For EfficientNet, we applied EfficientNet-B0–B7. From the experiments, EfficientNet-B2 is the superior, with the highest level of training accuracy of 99.90%, testing accuracy of 99.55%, precision of 99.50%, recall of 99.67%, and F1-Score of 99.58% with a training time of 15 minutes. The development of this automatic classification can assist radiologists in classifying brain tumor types more efficiently.

Author 1: Faiz Ainur Razi

Author 2: Alhadi Bustamam

Author 3: Arnida L. Latifah

Author 4: Shandar Ahmad

Keywords: Brain tumor; enhanced deep learning; MRI; multiclass; neuroimaging

Paper 121: Nature-Inspired Optimization for Virtual Machine Allocation in Cloud Computing: Current Methods and Future Directions

Abstract: An expanding range of services is offered by cloud data centers. The execution of application tasks is facilitated by assigning (VMs) Virtual Machines to (PMs) Physical Machines. Speaking of VM allocation in the cloud service center, two key factors are taken into consideration: quality of service (QoS) and energy consumption. The cloud service center aims to optimize these aspects while allocating VMs. On the other hand, cloud users have their priorities and focus on their specific requirements, particularly throughput and reliability. User requirements are considered by the cloud service center, resulting in VM allocation that meets QoS targets and optimizes energy consumption. Cloud service centers must, therefore, find a balance between QoS and energy efficiency while considering the user's requirements. To achieve this, various optimization algorithms and techniques must be employed. The objective is to find the best allocation of VMs to PMs. Due to the NP-hardness of the VM allocation problem, nature-inspired meta-heuristic algorithms have become commonly used to solve it. However, there are no comprehensive and in-depth review papers on this specific area. This paper aims to bridge a knowledge gap by providing an understanding of the significance of metaheuristic methods to address the VM allocation issue effectively. It not only highlights the role played by these algorithms but also examines the existing methods, provides comprehensive comparisons of strategies based on key parameters, and concludes with valuable recommendations for future research.

Author 1: Xiaoqing YANG

Keywords: Cloud computing; virtualization; virtual machine allocation; optimization

Paper 122: Investigating the Effectiveness of ChatGPT for Providing Personalized Learning Experience: A Case Study

Abstract: The demand for personalized learning experiences that cater to the unique needs of individual learners has increased with the emergence of data science. This paper investigates the potential use of ChatGPT, a generative AI tool, in providing personalized learning experiences for data science education, specifically focusing on Deep Learning. The paper presents a case study that applies the 5Es model to test personalized learning for students using ChatGPT. The study aims to answer the question of how educators can leverage ChatGPT in their pedagogy to enhance student learning, and whether ChatGPT can provide a better learning experience than traditional teaching methods. The paper also discusses the limitations faced during the study and the findings. The results suggest that ChatGPT can be a valuable resource for data science education, providing personalized and instant feedback to learners. However, ethical considerations such as the potential for biased or inaccurate responses and the need for transparency in AI-generated content should be carefully ad-dressed by educators. The study highlights ChatGPT’s potential as a research tool for data science educators to investigate the effectiveness of AI in personalized learning experiences. Overall, this paper contributes to the ongoing dialogue on the role of AI in data science education and provides insights into how educators can utilize ChatGPT to enhance student learning and engagement.

Author 1: Raneem N. Albdrani

Author 2: Amal A. Al-Shargabi

Keywords: Personalized learning; data science education; ChatGPT; generative AI

Paper 123: Securing Digital Data: A New Edge Detection and XOR Coding Approach for Imperceptible Image Steganography

Abstract: The rapid progress of digital devices and technol-ogy, coupled with the emergence of the internet has amplified the risks and perils associated with malicious attacks. Consequently, it becomes crucial to protect valuable information transmitted through the internet. Steganography is a tried-and-true technique for hiding information beneath digital content, such as pictures, texts, audio, and video. Various methodologies of image steganog-raphy have been developed recently. In image recognition, edge detection secures an image into well-defined areas. This paper introduces a novel image steganography algorithm with edge detection and XOR coding techniques. The proposed approach aims to conceal a confidential message within the spatial domain of the original image. In contrast to uniform regions, the Human Visual System (HVS) is less responsive to variations in the sharp areas; an edge detection algorithm is applied to identify edge pixels. Furthermore, to enhance the efficiency and reduce the embedding impact, XOR operation has been utilized to embed the secret message in the Least Significant Bit (LSB). According to the results of the experiments, the proposed method embeds confidential data without causing noticeable modifications to the stego image. The proposed method system produced imperceptible stego images with minimal embedding distortions compared to existing methods. Based on the results, the proposed approach outperforms the conventional methods regarding image distortion techniques. The PSNR values achieved by the proposed method are higher than the acceptable level.

Author 1: Hayat Al-Dmour

Keywords: Steganography; information hidings; bits modifica-tion; decoding algorithm; edge detection; canny edge detection; human visual system

Paper 124: Incorporating News Tags into Neural News Recommendation in Indonesian Language

Abstract: News recommendation system holds the potential to aid users in discovering articles that align with their interests, which is critical to alleviate user information overload. To generate effective news recommendations, one key capability is to accurately capture the contextual meaning of text in the news articles, since this is pivotal in acquiring useful repre-sentations for both news content and users. In this work, we examine the effectiveness of neural news recommendation with attentive multi-view learning (NAML) method to conduct a news recommendation task in the Indonesian language. We further propose to incorporate news tags, which at some levels may capture the important contextual meanings contained in the news articles, to improve the effectiveness of the NAML method in the Indonesian news recommendation system. Our results show that the NAML method leads to significant improvement (if not comparable) in the effectiveness of neural-based Indonesian news recommendations. Further incorporating news tags is shown to significantly increase the performance of the NAML method by 5.86% in terms of NDCG@5 metric.

Author 1: Maxalmina Satria Kahfi

Author 2: Evi Yulianti

Author 3: Alfan Farizki Wicaksono

Keywords: News recommendation; recommendation systems; news tags; user modeling

Paper 125: Telemedicine Adoption for Healthcare Delivery: A Systematic Review

Abstract: Telemedicine is the delivery of healthcare ser-vices using telecommunication and information technologies. The adoption of telemedicine has been promoted by advancements in technology, increased accessibility to the Internet, and the need for convenient and efficient healthcare delivery. Under-standing the theoretical foundations of telemedicine adoption among healthcare providers and patients is crucial for successful acceptance and utilization. This systematic review aims to explore the theoretical frameworks and models that have been widely utilized to understand telemedicine adoption among healthcare providers and patients. A systematic search was conducted across two popular electronic databases, resulting in the inclusion of 21 relevant studies. The selected studies were analyzed to identify the theoretical perspectives employed in telemedicine adoption research. The key findings reveal that the Technology Acceptance Model (TAM), the Unified Theory of Acceptance, and the Use of Technology (UTAUT) model are the most widely models used to illustrate the factors affecting telemedicine adoption among healthcare providers and patients through different countries and telemedicine contexts. Understanding these theoretical models is crucial for policymakers and healthcare professionals as it can provide insight into the key factors influencing the widespread adoption of telemedicine. This knowledge can serve as a guidance for crafting initiatives, and tailoring policies to promote the successful acceptance and utilization of telemedicine among providers and patients in diverse healthcare environments.

Author 1: Taif Ghiwaa

Author 2: Imran Khan

Author 3: Martin White

Author 4: Natalia Beloff

Keywords: Telemedicine; systematic review; technology accep-tance model; adoption; telehealth; healthcare provider; patient

Paper 126: Attention-based Cross-Modality Multiscale Fusion for Multispectral Pedestrian Detection

Abstract: Multispectral pedestrian detection has wide ap-plications in fields such as autonomous driving and intelli-gent surveillance. Mining complementary information between modalities is one of the most effective approaches to improve the performance of multispectral pedestrian detection. However, the inevitable introduction of redundant information between modalities during the fusion process leads to feature degradation. To address this challenge, we propose a multiscale differen-tial fusion algorithm that leverages complementary information between modalities to suppress feature degradation caused by noise propagation along the network. We compare our algorithm with other cross-modal fusion pedestrian detection algorithms on the LLVIP and cleaned KAIST datasets. Experimental results demonstrate that our algorithm outperforms others, particularly in nighttime scenes where our algorithm achieves a 7.28%improvement in recall rate compared to the baseline on the cleaned KAIST dataset.

Author 1: Zhou Hui

Keywords: Pedestrian detection; multispectral pedestrian detec-tion; attention mechanism; cross-modal fusion

Paper 127: Deep Learning-Powered Mobile App for Fast and Accurate COVID-19 Detection from Chest X-rays

Abstract: The COVID-19 pandemic has imposed significant challenges on healthcare systems globally, necessitating swift and precise screening methods to curb transmission. Traditional screening approaches are time-consuming and prone to errors, prompting the development of an innovative solution - a mobile application employing machine learning for automated COVID- 19 screening. This application harnesses computer vision and deep learning algorithms to analyze X-ray images, rapidly de-tecting virus-related symptoms. This solution aims to enhance the accuracy and speed of COVID-19 screening, particularly in resource-constrained or densely populated settings. The paper details the use of convolutional neural networks (CNNs) and transfer learning in diagnosing COVID-19 from chest X-rays, highlighting their efficacy in image classification. The trained model is deployed in a mobile application for real-world testing, aiming to aid healthcare professionals in the battle against the pandemic. The paper provides a comprehensive overview of the background, methodology, results, and the application’s architecture and functionalities, concluding with avenues for future research.

Author 1: Rahhal Errattahi

Author 2: Fatima Zahra Salmam

Author 3: Mohamed Lachgar

Author 4: Asmaa El Hannani

Author 5: Abdelhak Aqqal

Keywords: COVID-19 diagnosis; computer vision; deep learn-ing; X-ray images; mobile application

Paper 128: Explicit Knowledge Database Interface Model System Based on Natural Language Processing Techniques and Immersive Technologies

Abstract: This work is focused on the proposal and de-velopment of an interface system model, based on natural language processing, immersive technologies and natural user interfaces, for the interaction with Explicit Knowledge databases. Five phases were proposed: The user testing characterization, the establishment of the state of the art and the theoretical foundation, the design and development of software, the system implementation and the functional tests and evaluation of the usability of the interface model. In order to establish the user testing characterization and the corresponding theoretical frame-work, the expert guide on Knowledge Management and Virtual Reality was followed, based on the approach of Usability and Computer Ergonomics compatible with the ISO 9241 standard. The traditional interfaces and the proposal in this work were evaluated for each of the metrics defined by the ISO 9241 standard, considering the dimensions of effectiveness, efficiency and satisfaction. The statistical test of “T-Student” established that there is enough evidence to confirm the existence of the following significant differences: Effectiveness is lower using the proposed interface model; efficiency and satisfaction is higher using the proposed interface model. Based on the conducted tests, it can be established that the proposed interface model is superior to the traditional interface in terms of the “Efficiency and Satisfaction” dimensions and inferior in terms of “Effec-tiveness.” Consequently, it can be concluded that the scientific article exploration model using VR and NLP is superior to the traditional model.

Author 1: Luis Alfaro

Author 2: Claudia Rivera

Author 3: Jose Herrera

Author 4: Antonio Arroyo

Author 5: Lucy Delgado

Author 6: Elisa Castaneda

Keywords: Knowledge management; explicit knowledge databases; natural language processing; natural user interfaces; Immersive technologies

Paper 129: CESSO-HCRNN: A Hybrid CRNN With Chaotic Enriched SSO-based Improved Information Gain to Detect Zero-Day Attacks

Abstract: Hackers use the vulnerability before programmers have a chance to fix it, which is known as a zero-day attack. Zero-day attackers have a variety of abilities, including the ability to alter files, control machines, steal data, and install malware or adware. When a series of complex assaults uses one or more zero-day exploits, the result is a zero-day attack path. Timely assessment of zero-day threats might be enabled by early detection of zero-day attack pathways. To detect this zero-day attack, this paper introduced a Chaotic Enriched Salp Swarm Optimization (CESSO) with the help of a hybrid Convolutional Recursive Neural Network (HCRNN) is implemented. The input data is retrieved from two datasets called IDS 2018 Intrusion CSVs (CSE-CIC-IDS2018) and NSL-KDD. The data is pre-processed with the help of data cleaning and normalization. A unique hybrid feature selection method that is based on the CESSO and Information Gain(IG) is introduced. The CESSO is also used to improve the Recursive Neural Network (RNN) performance to produce an optimized RNN. The selected features are classified, and prediction is performed using the hybrid Convolutional Neural Network (CNN) with RNN called HCRNN. The implementation of the zero-day attack is performed using MATLAB software. The accuracy achieved for dataset 1 is 98.36%, and for dataset 2 is 97.14%.

Author 1: Dharani Kanta Roy

Author 2: Ripon Patgiri

Keywords: Hackers; vulnerability; zero-day attack; chaotic en-riched salp swarm optimization; data cleaning; normalization; and MATLAB software

Paper 130: Triggered Screen Restriction: Gamification Framework

Abstract: The prevalence of sedentary lifestyles is increasingly becoming a significant public health concern, with numerous health risks ranging from obesity to heart disease. Several gamified interventions have been employed to counter sedentary behavior by promoting physical activity. However, the existing approaches have yielded mixed results, making it crucial to explore new methodologies. While existing approaches have utilized gamification elements to encourage activity, they often need a comprehensive blend of psychological elements and advanced technology to drive a meaningful behavioral alteration. This paper introduces the Triggered Screen Restriction (TSR) framework, an interdisciplinary approach integrating behavioral psychology, gamification, and screen-time restriction technologies. The TSR framework aims to elevate gamified physical activity by leveraging the psychological Fear of Missing Out phenomenon, encouraging users to meet specific activity goals to unlock social media applications. The TSR framework presents a promising avenue for future research. The proposed framework’s unique approach is designed to motivate users to be more physically active. The proposed framework fills a literature gap in the current implementation of the gamified physical intervention. Further studies are needed to empirically validate the framework’s effectiveness and potential to contribute to the gamification ecosystem.

Author 1: Majed Hariri

Author 2: Richard Stone

Keywords: Gamification; physical activity; sedentary behavior; Triggered Screen Restriction (TSR) framework

Paper 131: A Particle Filter based Visual Object Tracking: A Systematic Review of Current Trends and Research Challenges

Abstract: Visual object tracking is a crucial research area in computer vision because it can simulate a dynamic environment with non-linear motions and multi-modal non-Gaussian noises. However, This paper presents an overview of the recent devel-opments in particle filter-based visual object tracking algorithms and discusses the pros and cons of particle filters, respectively. There are presentations of many different methodologies and algorithms in the research literature. The majority of visual object tracking research at present is on particle filters. In addition, the most advanced technique for visual object tracking has also been developed by combining the convolutional neural network (CNN) and the particle filter. The advantage of particle filters is that they can handle nonlinear models and non-Gaussian advancements, sequentially concentrating on the areas of the state space with higher densities, primarily parallelization, and simplicity of implementation. Despite this, it offers a robust framework for visual object tracking because it incorporates uncertainty and outperforms other filters like the Kalman filter, Kernelized correlation filter, optical filter, mean shift filter, and extended Kalman filter in recognition tests. In contrast, this study provided information on various particle filter features and classifiers.

Author 1: Md Abdul Awal

Author 2: Md Abu Rumman Refat

Author 3: Feroza Naznin

Author 4: Md Zahidul Islam

Keywords: Particle filter; visual object tracking; On-Gaussian noises; Kalman filter; CNN

Paper 132: Deep Speech Recognition System Based on AutoEncoder-GAN for Biometric Access Control

Abstract: Speech recognition-based biometric access control systems are promising solutions that have resolved many is-sues related to security and convenience. Speech recognition, as a biometric modality, offers unique advantages such as user-friendliness and non-intrusiveness, etc. However, developing robust and accurate speaker identification and authentication systems pose challenges due to variations in speech patterns and environmental factors. Integrating deep learning techniques, especially AutoEncoder and Generative Adversarial Network models, has shown promising results in addressing these chal-lenges. This article presents a novel approach based on the combination of two deep learning models, namely, AE and GAN for speech recognition-based biometric access control. In the model architecture, the AutoEncoder takes the MFCC coefficients as input, and the encoder converts the latter to the latent space, whereas the decoder reconstructs the data. Then, speech features extracted from the latent space are used in the GAN generator to generate additional speech data. The discriminator network has a dual role, serving as both a feature extractor and a classifier. The first extracts relevant features from generated samples, while the latter distinguishes between generated and authentic samples that come from AutoEncoder. This strategy outperforms DNN and LSTM models on VoxCeleb 2, LibriSpeech, and Aishell- 1 datasets. The models are trained to minimize Mean Squared Error (MSE) for both the generator and discriminator, aiming at achieving highly realistic datasets and a robust, interpretable model. This approach addresses challenges in feature extraction, data augmentation, realistic biometric samples generation, data variability handling, and data generalization enhancement, pro-viding therefore, a comprehensive solution.

Author 1: Oussama Mounnan

Author 2: Otman Manad

Author 3: Abdelkrim El Mouatasim

Author 4: Larbi Boubchir

Author 5: Boubaker Daachi

Keywords: Speaker identification; speech recognition; biomet-ric access control; authentication; verification

Paper 133: Estimation of Hazardous Environments Through Speech and Ambient Noise Analysis

Abstract: In recent years, significant attention has been di-rected towards the development of artificial empathy within the engineering academic community. Replicating artificial empathy necessitates the capability of agents to discern human emotions and comprehend environmental risks. Analyzing acoustic data in real environments offers a higher level of non-invasive pri-vacy compared to video and camera data, limiting the agent’s understanding to specific patterns. However, current studies are negatively affected by subjective inferences from real data, which can result in inaccurate predictions, leading to both false positives and negatives, especially when contextual data and human speech are involved. This paper work proposes the estimation of a dangerous environment in accordance with the emotional speech and additional ambient noises. In this approach we implement a variational autoencoder model in conjunction with a classifier for training the classification task. Additional regularization techniques are applied to bridge the gap between the original training data and the expected data. The classifier utilizes feature data generated by the variational autoencoder to extract class patterns and determine whether the environment is hazardous. Emotional speech is classified as angry, sad, or scared emotions, contributing to the classification of danger, while happy, calm, and neutral emotions are considered safe. Various ambient noise types, including gunfire and broken glass, are categorized as dangerous, while real-life indoor noises like cooking, eating, and movements are considered safe.

Author 1: Andrea Veronica Porco

Author 2: Kang Dongshik

Keywords: Dangerous environment detection; speech analysis; acoustic audio analysis; ambient noises; variational autoencoder model; empathetic systems

Paper 134: D2-Net: Dilated Contextual Transformer and Depth-wise Separable Deconvolution for Remote Sensing Imagery Detection

Abstract: Remote sensing-based object detection faces chal-lenges in arbitrary orientations, complex backgrounds, dense distributions, and large aspect ratios. Considering these issues, this paper introduces a novel method called D2-Net, which incorporates a transformer structure into a convolutional neural network. First, a new feature extraction module called dilated contextual transformer block is designed to minimize the loss of object information due to complex backgrounds and dense tar-gets. In addition, an efficient approach using depth-wise separable deconvolution as an up-sampling method is developed to recover lost feature information effectively. Finally, the circular smooth label is incorporated to compute the angular loss to complete the rotated detection of remote sensing images. Experimental evaluations are conducted on the DOTA and HRSC2016 datasets. On the DOTA dataset, the proposed method achieves 79.2%and 78.00% accuracy in horizontal and rotated object detection, respectively; it achieves 94.00% accuracy in the rotated detection of the HRSC2016 dataset. The proposed model shows a significant performance improvement over other comparative models on the dataset, which verifies the effectiveness of our proposed approach.

Author 1: Huaping Zhou

Author 2: Qi Zhao

Author 3: Kelei Sun

Keywords: YOLOv7; dilated contextual transformer; depth-wise separable deconvolution; circular smooth label; remote sensing

Paper 135: Semantic Embeddings for Arabic Retrieval Augmented Generation (ARAG)

Abstract: In recent times, Retrieval Augmented Generation (RAG) models have garnered considerable attention, primarily due to the impressive capabilities exhibited by Large Language Models (LLMs). Nevertheless, the Arabic language, despite its significance and widespread use, has received relatively less research emphasis in this field. A critical element within RAG systems is the Information Retrieval component, and at its core lies the vector embedding process commonly referred to as “semantic embedding”. This study encompasses an array of multilingual semantic embedding models, intending to enhance the model’s ability to comprehend and generate Arabic text effec-tively. We conducted an extensive evaluation of the performance of ten cutting-edge Multilingual Semantic embedding models, employing a publicly available ARCD dataset as a benchmark and assessing their performance using the average Recall@k metric. The results showed that the Microsoft E5 sentence embedding model outperformed all other models on the ARCD dataset, with Recall@10 exceeding 90%.

Author 1: Hazem Abdelazim

Author 2: Mohamed Tharwat

Author 3: Ammar Mohamed

Keywords: Arabic NLP; large language models; retrieval aug-mented generation; semantic embedding

Paper 136: Elevating Android Privacy: A Blockchain-Powered Paradigm for Secure Data Management

Abstract: The significance of medical test records in diagnos-ing and treating illnesses cannot be overstated. These records serve as the foundation upon which medical professionals craft precise treatment strategies tailored to a patient’s unique health condition and ailment. However, in several developing nations, such as Vietnam, a concerning trend persists: medical test records predominantly exist in vulnerable paper format, entrusted to patients for safekeeping. When patients transition between health-care facilities, the responsibility of carrying these paper-based medical histories rests with them, introducing a significant risk factor due to the inherent fragility of paper documents, which can be easily damaged by fire or water. The loss of these crucial records can lead to severe disruptions in the diagnostic and therapeutic journey of patients, potentially compromising their well-being. Despite the emergence of various alternatives to address this vulnerability, Vietnam faces multifaceted challenges. These challenges encompass low technological literacy among patients and substantial infrastructural limitations. In response to these pressing issues, this study endeavors to harness the transformative potential of blockchain technology, smart con-tracts, and Non-Fungible Tokens (NFTs) to effectively mitigate the drawbacks associated with paper-based medical test records. Our comprehensive approach includes meticulous cataloging of current hospital practices, the introduction of a purpose-built blueprint for decentralized record sharing, the proposal of an innovative NFT-backed authentication model, the development of a practical proof-of-concept, and comprehensive platform testing. Through these efforts, we aim to revolutionize the management of medical test records in Vietnam, enhancing accessibility, security, and reliability for both patients and healthcare providers.

Author 1: Bang Khanh Le

Author 2: Ngan Thi Kim Nguyen

Author 3: Khiem Gia Huynh

Author 4: Phuc Trong Nguyen

Author 5: Anh The Nguyen

Author 6: Khoa Dand Tran

Author 7: Trung Hoang Tuan Phan

Keywords: Medical test result; blockchain; smart contract; NFT; Ethereum; Fantom; Polygon; Binance smart chain

Paper 137: A Deep Transfer Learning Approach for Accurate Dragon Fruit Ripeness Classification and Visual Explanation using Grad-CAM

Abstract: Dragon fruit, known for its rich antioxidant content and low-calorie attributes, has garnered significant attention as a health-promoting fruit. Its economic value has also surged due to increasing consumer demand and its potential as an export commodity in various regions. The classification of dragon fruit ripeness is a pivotal task in ensuring product quality and minimizing post-harvest losses. This research article presents a comprehensive study on the classification of ripe and unripe dragon fruits (Hylocereus spp) using the Densenet201 model through three distinct approaches: as a classifier, feature extrac-tor, and fine-tuner. To explain the outcomes of the image clas-sification model and thereby enhance its performance, optimiza-tion, and reliability, this study employs advanced visualization techniques. Specifically, it utilizes Grad-CAM (Gradient-weighted Class Activation Mapping) and Guided Grad-CAM techniques. These techniques offer insights into the model’s decision-making process and pinpoint regions of interest within the images. This approach empowers researchers to iteratively validate the model’s accuracy and enhance its performance. The utilization of Densenet201 as a classifier, feature extractor, and fine-tuner, coupled with the insights from Grad-Cam and Guided Grad-Cam, presents a holistic approach to enhancing dragon fruit ripeness classification. The findings contribute to the broader discourse on agricultural technology, image analysis, and the optimization of classification models.

Author 1: Hoang-Tu Vo

Author 2: Nhon Nguyen Thien

Author 3: Kheo Chau Mui

Keywords: Dragon fruit classification; ripeness classification; densenet201 model; Grad-CAM visualization; guided grad-CAM; visual interpretation; Explainable AI; XAI; deep learning; pre-trained models; model fine-tuning; transfer learning

Paper 138: Identification of Air-Writing Tamil Alphabetical Vowel Characters

Abstract: In recent years, there has been a lot of focus on gesture recognition because of its potential as a means of communication for cutting-edge gadgets. As a special category of gesture recognition, air-writing is the practice of forming letters or words in the air using one’s fingers or the move-ments of one’s hands. The primary objective of this study is to propose a classification framework with feature extraction techniques to enhance the recognition of vowel characters in the Tamil language. The data collection and classification procedure involved a set of 12 distinct letters. A methodology has been developed to facilitate the analysis of various configurations for the purpose of evaluation. To get useful features from the 2-second time window data segments, this study uses a one-dimensional convolutional neural network (1D CNN). In our approach, we employ five machine-learning methods to conduct our evaluation. These methods include Naive Bayes, Random Forest, K-Nearest Neighbor, Support Vector Machine, and Decision Tree. The classification algorithms are considered to be superior based on the results obtained from our dataset in this experiment. The results of the tests show that the suggested K-nearest neighbors (KNN) algorithm works very well when used with a k-1 and 0.6:0.4 split ratio for training and testing. Specifically, the KNN model achieved an accuracy rate of 91.67%. The present study builds upon previous research by utilizing applications that have been employed in prior studies. However, a unique aspect of our system is the integration of cutting-edge technology, which utilizes collected sensor data to classify the characters. The examination of the window size has the potential to enhance accuracy and performance.

Author 1: Rukshani Puvanendran

Author 2: Vijayanathan Senthooran

Keywords: Air-writing; Tamil alphabetical vowel; convolutional neural network; feature extraction; machine learning

Paper 139: Emotional Speech Transfer on Demand based on Contextual Information and Generative Models: A Case Study

Abstract: The automated generation of speech audio that closely resembles human emotional speech has garnered signif-icant attention from the society and the engineering academia. This attention is due to its diverse applications, including au-diobooks, podcasts, and the development of empathetic home assistants. In the scope of this study, it is introduced a novel approach to emotional speech transfer utilizing generative models and a selected emotional target desired for the output speech. The natural speech has been extended with contextual information data related with emotional speech cues. The generative models used for pursuing this task are a variational autoencoder model and a conditional generative adversarial network model. In this case study, an input voice audio, a desired utterance, and user-selected emotional cues, are used to produce emotionally expressive speech audio, transferring an ordinary speech audio with added contextual cues, into a happy emotional speech audio by a variational autoencoder model. The model try to reproduce in the ordinary speech, the emotion present in the emotional contextual cues used for training. The results show that, the proposed unsupervised VAE model with custom dataset for generating emotional data reach an MSE lower than 0.010 and an SSIM almost reaching the 0.70, while most of the values are greater than 0.60, respect to the input data and the generated data. CGAN and VAE models when generating new emotional data on demand, show a certain degree of success in the evaluation of an emotion classifier that determines the similarity with real emotional audios.

Author 1: Andrea Veronica Porco

Author 2: Kang Dongshik

Keywords: Emotion transfer; contextual information; speech processing; generative models; variational autoencoder; conditional generative adversarial networks; empathetic systems

Paper 140: Imbalance Node Classification with Graph Neural Networks (GNN): A Study on a Twitter Dataset

Abstract: Social networks produce a large volume of infor-mation, a part of which is fake. Social media platforms do a good job in moderating content and banning fake news spreaders, but a proactive solution is more desirable especially during global threats like COVID-19 pandemic and war. A proactive solution would be to ban users who spread fake news before they become important spreaders. In this paper we propose to model user’s interactions in a social media platform as a graph and then evaluate state of the art (SOTA) graph neural networks (GNN) that can classify users’ (nodes) profiles as being suspended or not. As with other real world data, we are faced with the imbalanced data problem and we evaluate different algorithms that try to fix this issue. Data used for this study were collected from X (Twitter) by using Twitter API 1.1 from November 2021 to July 2022 with the focus to collect information spread through tweets about vaccines. The aim of this paper is to evaluate if current models can deal with real world imbalanced data.

Author 1: Alda Kika

Author 2: Arber Ceni

Author 3: Denada Collaku

Author 4: Emiranda Loka

Author 5: Ledia Bozo

Author 6: Klesti Hoxha

Keywords: GNN; imbalanced data; Twitter; social networks; GCN; GraphSage; GAT; GraphSMOTE; ReNode

Paper 141: Preventing Cyberbullying on Social Networks with Spanish Parental Control NLP System

Abstract: The boom in social networks and digital communi-cation has given place to innovative forms of social interaction. However, it has also made possible new forms of harassment of others anonymously and without repercussions. Such is the case of cyberbullying, an increasingly common problem, especially among young people. Its effects on individuals can be devastating, ranging from anxiety and depression to social isolation and low self-esteem. Furthermore, there is a wide variety of applications called parental control, which allow parents to show, the pages the child or adolescent has accessed, know how often the child or adolescent accesses them, and control the time spent on social networks or other entertainment platforms. Therefore, the present research aimed to analyze, design, and implement an intelligent application based on data mining algorithms and the Latent Semantic Analysis (LSA) method for the presumed detection of cyberbullying in social networks in adolescents. The methodological process of the study was carried out following the fundamentals of applied research with a qualitative-quantitative descriptive, and cross-sectional approach. As a result, a multi-platform application was obtained that alerts about suspected bullying to parents or guardians. For the validation of the application, the technique of expert judgment was applied. Also, the process of obtaining negative and positive text similarity was performed based on cosine similarity. In the analysis of Twitter accounts, values of 46% with negative texts and 6.71% with positive texts are obtained, which allows inferring that this is a presumed case of cyberbullying in this account.

Author 1: Gabriel A. Leon-Paredes

Author 2: Omar G. Bravo-Quezada

Author 3: Pedro P. Bermeo-Aguaysa

Author 4: Maria J. Pelaez-Currillo

Author 5: Ledys L. Jimenez-Gonzalez

Keywords: Cyberbullying; control parental system; natural lan-guage processing; Spanish cyberbullying prevention system

Paper 142: Mukh-Oboyob: Stable Diffusion and BanglaBERT enhanced Bangla Text-to-Face Synthesis

Abstract: Facial image generation from textual generation is one of the most complicated tasks within the broader topic of Text-to-Image (TTI) synthesis. It is relevant in several fields of scientific research, cartoon and animation development, online marketing, game development, etc. There have been extensive studies on Text-to-Face (TTF) synthesis in the English language. However, the amount of relevant existing work in Bangla is limited and not comprehensive. As the TTF field is not vastly prospected for Bangla language, the objective of this study sets forth to explore the possibilities in the field of Bangla Natural Language Processing and Computer Vision. In this paper, a novel system for generating highly detailed facial images from textual descriptions in the Bangla language is proposed. The proposed system named Mukh-Oboyob consists of two essential components: a pre-trained language model, BanglaBERT, and Stable Diffusion. BanglaBERT, a transformer-based pre-trained text encoder, is a language model used to transform Bangla sentences into vector representations. Stable Diffusion is used by Mukh-Oboyob to generate facial images utilizing the text embedding of the Bangla sentences. Moreover, the work uti-lizes CelebA Bangla, a modified version of the CelebA dataset consisting of face images, Bangla facial attributes, and Bangla text descriptions to develop and train the proposed system. This paper establishes a system for image synthesis with excellent performance and detailed image outcomes, as evidenced by a comprehensive analysis incorporating both qualitative and quantitative measures, leading to the system under consideration achieving an impressive FID score of 34.6828 and an LPIPS score of 0.4541.

Author 1: Aloke Kumar Saha

Author 2: Noor Mairukh Khan Arnob

Author 3: Nakiba Nuren Rahman

Author 4: Maria Haque

Author 5: Shah Murtaza Rashid Al Masud

Author 6: Rashik Rahman

Keywords: Bangla text-to-face synthesis; Natural Language Processing (NLP); Bangla NLP; Computer Vision (CV); Generative Model; stable diffusion; BanglaBERT

Paper 143: Generate Adversarial Attack on Graph Neural Network using K-Means Clustering and Class Activation Mapping

Abstract: Graph Neural Networks (GNNs) have emerged as powerful tools for analyzing complex structured data, including social networks, biological networks, and recommendation sys-tems. However, their susceptibility to adversarial attacks poses a significant challenge, especially in critical tasks such as node classification and link prediction. Adversarial attacks on GNNs can introduce harmful input graphs, leading to biased model predictions and compromising the integrity of the network. We propose a novel adversarial attack method that leverages the combination of K-Means clustering and Class Activation Mapping (CAM) to conduct subtle yet effective attacks against GNNs. The clustering algorithm identifies critical nodes within the graph, whose perturbations are likely to have a substantial impact on model performance. Additionally, CAM highlights regions of the graph that significantly influence GNN predictions, enabling more targeted and efficient attacks. We assess the efficacy of state-of-the-art GNN defenses against our proposed attack, underscoring the pressing need for robust defense mechanisms. Our study focuses on countering attacks on GNN networks by utilizing K-Means clustering and CAM to enhance the effectiveness and efficiency of the adversarial strat-egy. Through our observations, we emphasize the necessity for stronger security measures to safeguard GNN-based applications, particularly in sensitive environments.Furthermore, our research highlights the importance of developing robust GNNs that can withstand adversarial attacks, ensuring the reliability and trustworthiness of these models in critical applications. Strengthening the robustness of GNNs against adversarial manipulation is crucial for maintaining the se-curity and integrity of systems that heavily rely on these advanced analytical tools. Our findings underscore the ongoing efforts required to fortify GNN-based applications, urging the research community and practitioners to collaborate in developing and implementing more robust security measures for these powerful neural network models.

Author 1: Ganesh Ingle

Author 2: Sanjesh Pawale

Keywords: Graph Neural Networks; adversarial attacks; K-Means clustering; class activation mapping; robustness; defense mechanisms

Paper 144: A Comprehensive Review of Deep Learning Approaches for Animal Detection on Video Data

Abstract: Integrating deep learning techniques into computer vision application has ushered in a new era of automated analysis and interpretation of visual data. In recent years, a surge of interest has been witnessed in applying these methodologies towards detecting animals in video streams, promising transformative impacts on diverse fields such as ecology and agriculture. This paper presents an extensive and meticulous review of the latest deep-learning approaches employed for animal detection in video data. This study looks closely at ways to detect animals in videos using deep learning. This study explores various Deep learning methods for detecting many animals in multiple environments. The analysis also pays close attention to preparing the data, picking out important features, and reusing what has been learned from one task to help with another. In addition to highlighting successful methodologies, this review addresses the challenges and limitations inherent in these approaches issues such as limited data availability and adapting to technological advancements present significant hurdles. Recognising and understanding these challenges is crucial in shaping the future focus of research endeavours. Thus, this comprehensive review is an indispensable tool for anyone keen on employing these potent computer methods for animal detection in videos. It takes the latest ideas and shows where study can explore further to improve them. Furthermore, this comprehensive review has demonstrated that a more sustainable and balanced relationship between humans and animals can be achieved by harnessing the power of deep learning in animal detection. This research contributes to computer vision and holds immense promise in safeguarding biodiversity and promoting responsible land use practices, especially within agricultural domains. The insights from this study propel us towards a future where advanced technology and ecological harmony go hand in hand, ultimately benefiting both humans and the animal kingdom. The survey aims to provide a comprehensive overview of the cutting-edge developments in applying deep learning models for animal detection through cameras by elucidating the significance of these techniques in advancing the accuracy and efficiency of animal detection processes.

Author 1: Prashanth Kumar

Author 2: Suhuai Luo

Author 3: Kamran Shaukat

Keywords: Machine learning; deep learning; animal detection; convolutional neural networks; video-based; deep learning models

Paper 145: Studying the Security and Privacy Issues of Big Data in the Saudi Medical Sector

Abstract: In today’s era of Big Data, with the integration of data from various systems, devices, and machines used by healthcare service providers, health insurance companies, and their sub-sectors, maintaining privacy and security has become crucial. It is important to uphold the confidentiality and security of data exchanged between data service providers and insurance companies as required by law. The purpose of this paper is to focus on addressing the security and privacy issues associated with healthcare data, particularly concerning medical data in both in- transit and at-rest modes. We aim to provide a proposed solution to enhance data security and maximize privacy protection.

Author 1: Ramy Elnaghy

Author 2: Hazem M. El-Bakry

Keywords: Security; privacy; healthcare; medical data; big data

Paper 146: A Novel Deep Learning-Assisted SVD-based Method for Medical Image Watermarking

Abstract: In the present era, the administration of medical images faces various security challenges that necessitate the authentication of image source and origin for accurate patient identification. With the increasing exchange of medical images between hospitals to facilitate informed decision-making, the adoption of digital watermarking techniques has emerged as an efficient solution to address the imperceptibility and robustness requirements in medical imaging watermarking. This research work introduces a technically advanced approach that combines singular value decomposition (SVD) watermarking with deep learning segmentation models to enhance the security of medical image sharing and transfer. The primary objective is to seamlessly integrate the watermark while minimizing distortion to preserve critical medical information within the image. The proposed methodology involves utilizing a ResNet-based U-Net segmentation model to segment X-Ray radiographs into the Region of Interest (ROI) and the Region of Non-Interest (RONI). The watermark data is then encoded into the ROI using singular value decomposition. Subsequently, the ROI and RONI are merged to reconstruct the complete image, preserving its original identity. Additionally, XOR encryption is applied to the watermarked image to enhance data integrity and copyright protection. On the other side of the methodology, the reconstructed image is once again separated into ROI and RONI. The ROI is decoded to recover the original transferred content. To assess the efficacy of the proposed method, a publicly available X-Ray radiograph dataset is employed, and evaluation metrics demonstrate an impressive segmentation accuracy of 98.27%. The proposed approach ensures information integrity, patient confidentiality during data sharing, and robustness against various conventional attacks, demonstrating its effectiveness in the field of medical image watermarking.

Author 1: Saima Kanwal

Author 2: Feng Tao

Author 3: Rizwan Taj

Keywords: Singular value decomposition; medical image watermarking; digital watermarking; deep learning