IJACSA Volume 14 Issue 5 - thesai.org

IJACSA Volume 14 Issue 5

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: Spatio-Temporal Features based Human Action Recognition using Convolutional Long Short-Term Deep Neural Network

Abstract: Recognition of human intention is crucial and challenging due to subtle motion patterns of a series of action evolutions. Understanding of human actions is the foundation of many applications, i.e., human robot interaction, smart video monitoring and autonomous driving etc. Existing deep learning methods use either spatial or temporal features during training. This research focuses on developing a lightweight method using both spatial and temporal features to predict human intention correctly. This research proposes Convolutional Long Short-Term Deep Network (CLSTDN) consists of Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN). CNN uses Inception-ResNet-v2 to classify object specific class categories by extracting spatial features and RNN uses Long Short-Term Memory (LSTM) for final prediction based on temporal features. Proposed method was validated on four challenging benchmark dataset, i.e., UCF Sports, UCF-11, KTH and UCF-50. Performance of the proposed method was evaluated using seven performance metrics, i.e., accuracy, precision, recall, f-measure, error rate, loss and confusion matrix. Proposed method showed better results comparing with existing research results. Proposed method is expected to encourage researchers to use in future for real time implications to predict human intentions more robustly.

Author 1: A F M Saifuddin Saif

Author 2: Ebisa D. Wollega

Author 3: Sylvester A. Kalevela

Keywords: Convolutional neural network; recurrent neural network; long short-term memory; human action recognition

Paper 2: Usability and Security of Knowledge-based Authentication Systems: A State-of-the-Art Review

Abstract: Knowledge-based passwords are still the most dominant authentication method for securing digital platforms and services, in spite of the emergence of alternative systems such as token-based and biometric systems. This method has remained the most popular one mostly because of its usability, compatibility, affordability of implementation, and user familiarity. However, the main challenge of knowledge-based password schemes lies in creating passwords that provide a balance between memorability and security. This research aimed to compare various knowledge-based schemes in order to establish a strategy that provided high memorability and resilience to most cyberattacks. The overview of this research identifies areas of knowledge-based passwords for further research and enhances the methodology that helps to offer insight into usable, secure, and sustainable authentication approaches. Future work has been recommended to explore the major features and drawbacks of recognition-based textual passwords because this method provides the usability and security benefits of graphical passwords with the familiarity of textual passwords.

Author 1: Hassan Wasfi

Author 2: Richard Stone

Keywords: Knowledge-based authentication; recognition; re-call; usability; security; memorability

Paper 3: Super-Resolution of Brain MRI via U-Net Architecture

Abstract: This paper proposes a U-Net-based deep learning architecture for the task of super-resolution of lower resolution brain magnetic resonance images (MRI). The proposed system, called MRI-Net, is designed to learn the mapping between low-resolution and high-resolution MRI images. The system is trained using 50-800 2D MRI scans, depending on the architecture, and is evaluated using peak signal-to-noise ratio (PSNR) metrics on 10 randomly selected images. The proposed U-Net architecture outperforms current state-of-the-art networks in terms of PSNR when evaluated with a 3 x 3 resolution downsampling index. The system's ability to super-resolve MRI scans has the potential to enable physicians to detect pathologies better and perform a wider range of applications. The symmetrical downsampling pipeline used in this study allows for generically representing low-resolution MRI scans to highlight proof of concept for the U-Net-based approach. The system is implemented on PyTorch 1.9.0 with NVIDIA GPU processing to speed up training time. U-Net is a promising tool for medical applications in MRI, which can provide accurate and high-quality images for better diagnoses and treatment plans. The proposed approach has the potential to reduce the costs associated with high-resolution MRI scans by providing a solution for enhancing the image quality of low-resolution scans.

Author 1: Aryan Kalluvila

Keywords: MRI; U-Net; Super-Resolution; PyTorch; SRCNN; SR-GAN; Deep Learning; GPU

Paper 4: Sentiment Analysis on COVID-19 Vaccine Tweets using Machine Learning and Deep Learning Algorithms

Abstract: One of the main functions of NLP (Natural Language Processing) is to analyze a sentiment or opinion of the text considered. In this research the objective is to analyze the sentiment in the form of tweets towards the Covid-19 vaccination. In this study, the collected tweets are in the form of a dataset from Kaggle that have been categorized into positive and negative depending on the polarity of the sentiment in that tweet, to visualize the overall situation. The reviews are translated into vector representations using various techniques, including Bag-Of-Words and TF-IDF to ensure the best result. Machine learning algorithms like Logistic Regression, Naïve Bayes, Support Vector Machine (SVM) and others, and Deep Learning algorithms like LSTM and Bert were used to train the predictive models. The performance metrics used to test the performance of the models show that Support Vector Machine (SVM) achieved the highest accuracy of 88.7989% among the machine learning models. Compared to the related research papers the highest accuracy obtained using LSTM is 90.59 % and our model has predicted with the highest accuracy of 90.42% using BERT techniques.

Author 1: Tarun Jain

Author 2: Vivek Kumar Verma

Author 3: Akhilesh Kumar Sharma

Author 4: Bhavna Saini

Author 5: Nishant Purohit

Author 6: Bhavika

Author 7: Hairulnizam Mahdin

Author 8: Masitah Ahmad

Author 9: Rozanawati Darman

Author 10: Su-Cheng Haw

Author 11: Shazlyn Milleana Shaharudin

Author 12: Mohammad Syafwan Arshad

Keywords: Covid-19 vaccine; sentiment analysis; machine learning; deep learning; natural language processing

Paper 5: An Enhanced SVM Model for Implicit Aspect Identification in Sentiment Analysis

Abstract: Opinion Mining or Sentiment Analysis (SA) is a key component of E-commerce applications where a vast number of reviews are generated by customers. SA operates on aspect level where the views are expressed on a specific aspect of a product and have a big influence on the customers’ choices and businesses’ reputation. Aspect Based Sentiment Analysis (ABSA) is the task of categorizing text by aspect and identifying the sentiment attributed to it. Implicit Aspect Identification (IAI) is a subtask of ABSA. This paper aims to empirically investigate how external knowledge (e.g. WordNet) is integrated into SVM model to address some of its intrinsic issues when dealing with classification. To achieve this research goal, we propose an approach to improve Support Vector Machines (SVM) model to deal with IAI. Using WordNet (WN) semantic relations, we suggest an enhancement for the SVM kernel computation. Experiments are conducted on three benchmark datasets of products, laptops, and restaurant reviews. The effects of our approach are examined and analyzed according to three criteria: (i) kernel function used, (ii) different experimental settings, and (iii) SVM behavior towards Overfitting and Underfitting. The research finding of our work is that the integration of external knowledge (e.g. WordNet) is experimentally proved to be significantly helpful to SVM classification for IAI and especially for addressing Overfitting and Underfitting that are considered as two of the main structural SVM issues. The empirical results demonstrate that our approach helps SVM (i) improve its performance for the three considered kernels and under different experimental settings, and (ii) deal better with Overfitting and Underfitting.

Author 1: Halima Benarafa

Author 2: Mohammed Benkhalifa

Author 3: Moulay Akhloufi

Keywords: Implicit aspect-based sentiment analysis; machine learning; supervised approaches; support vector machines; wordnet; lesk algorithm

Paper 6: Ethereum Cryptocurrency Entry Point and Trend Prediction using Bitcoin Correlation and Multiple Data Combination

Abstract: Deep learning methods have achieved significant success in various applications, including trend signal prediction in financial markets. However, most existing approaches only utilize price action data. In this paper, we propose a novel system that incorporates multiple data sources and market correlations to predict the trend signal of Ethereum cryptocurrency. We conduct experiments to investigate the relationship between price action, candlestick patterns, and Ethereum-Bitcoin correlation, aiming to achieve highly accurate trend signal predictions. We evaluate and compare two different training strategies for Convolutional Neural Networks (CNNs), one based on transfer learning and the other on training from scratch. Our proposed 1-Dimensional CNN (1DCNN) model can also identify inflection points in price trends during specific periods through the analysis of statistical indicators. We demonstrate that our model produces more reliable predictions when utilizing multiple data representations. Our experiments show that by combining different types of data, it is possible to accurately identify both inflection points and trend signals with an accuracy of 98%.

Author 1: Abdellah EL ZAAR

Author 2: Nabil BENAYA

Author 3: Hicham EL MOUBTAHIJ

Author 4: Toufik BAKIR

Author 5: Amine MANSOURI

Author 6: Abderrahim EL ALLATI

Keywords: Deep learning; cryptocurrency; bitcoin trend prediction; price action; convolutional neural network; transfer learning

Paper 7: Security in the IoT: State-of-the-Art, Issues, Solutions, and Challenges

Abstract: Now-a-days, the Internet of Things (IoT) has enormous potential and growth impact due to the technological revolution and the spread and appearance of events. It has received considerable attention from researchers and is considered the future of the Internet; however, according to Cisco Inc. reports, the IoT will be crucial in transforming our standards of living, as well as our corporate and commercial models. By 2023, the number of devices connected to IP networks will reach more than three times the population of the entire world. In addition, there will be 5.3 billion Internet users worldwide, representing 66% of the world's population, up from 3.9 billion in 2018. IoT enables billions of devices and services to connect to each other and exchange information; however, most of these IoT devices can be easily compromised and are subject to various security attacks. In this article, we present and discuss the main IoT security issues, categorizing them according to the IoT layer architecture and the protocols used for networking. In the following, we describe the security requirements as well as the current attacks and methods with adequate solutions and architecture for avoiding these issues and security breaches.

Author 1: Ahmed SRHIR

Author 2: Tomader MAZRI

Author 3: Mohammed BENBRAHIM

Keywords: Internet of things (IoT); IoT security; IoT protocols; security issues in IoT; network security; data security

Paper 8: A Mobile App for the Identification of Flowers Using Deep Learning

Abstract: Flowers are admired and used by people all around the world for their fragrance, religious significance, and medicinal capabilities. The accurate taxonomy of these flower species is critical for biodiversity conservation and research. Non-experts typically need to spend a lot of time examining botanical guides in order to accurately identify a flower, which can be challenging and time-consuming. In this study, an innovative mobile application named FloralCam has been developed for the identification of flower species that are commonly found in Mauritius. Our dataset, named FlowerNet, was collected using a smartphone in a natural environment setting and consists of 11660 images, with 110 images for each of the 106 flower species. Seventy percent of the data was used for training, twenty percent for validation and the remaining ten percent for testing. Using the approach of transfer learning, pre-trained convolutional neural networks (CNNs) such as the InceptionV3, MobileNetV2 and ResNet50V2 were fine tuned on the custom dataset created. The best performance was achieved with the fine tuned MobileNetV2 model with accuracy 99.74% and prediction time 0.09 seconds. The best model was then converted to TensorFlow Lite format and integrated in a mobile application which was built using Flutter. Furthermore, the models were also tested on the benchmark Oxford 102 dataset and MobileNetV2 obtained the highest classification accuracy of 95.90%. The mobile application, the dataset and the deep learning models developed can be used to support future research in the field of flower recognition.

Author 1: Gandhinee Rajkomar

Author 2: Sameerchand Pudaruth

Keywords: Flowers; deep learning; mobile application; Mauritius

Paper 9: A Study of Prediction of Airline Stock Price through Oil Price with Long Short-Term Memory Model

Abstract: This study aims to present a model that predicts the stock price of an airline by setting the economic and technical information of oil as features and taking advantage of the LSTM method. In this study, oil price data for about seven years from January 4, 2016, to April 14, 2023, were collected through FinanceDataReader. The collected data is a total of 1,833 days of AA stock price data. The price data consists of six categories: Date, Open, High, Low, Close, Volume, and Change (price is based on dollars). Data is stored every 24 hours, so it was judged to be most suitable for short-term price prediction (24 hours later) to be conducted in this study. In this paper, normalized closing price data was trained for 50 epochs. As a result of learning, the loss value converged close to 0. The MSE measured by the accuracy of the model shows a result of 0.00049. The significance of this study is as follows. First, it is meaningful in that it can present indicators such as more sophisticated predictions and risk management to airline companies. Oil price as our selected feature can compensate for the poor performance of a simple model and its limitations on overfitting.

Author 1: Jae Won Choi

Author 2: Youngkeun Choi

Keywords: Stock price prediction; airline; oil; long short-term memory

Paper 10: Method for Ad-hoc Blockchain of Wireless Mesh Networking with Agent and Initiate Nodes

Abstract: Method for Ad-Hoc blockchain of wireless mesh networking with agent and initiate nodes is proposed. Minimizing the number of hops and maintaining connectivity of mobile terminals are concerns. Through simulation studies, it is found that increasing number of initiator nodes caused nodes to route a large number of messages. Thus, these nodes will die out quickly, causing the energy required to get the remaining messages to increase and more nodes to die. This will create a cascading effect that will shorten system lifetime. Multi-hop routing, however, imply high packet overhead, (more nodes in the network means more hops will be available). The packet overhead of the multi-hop routing is extremely high compared to single path routing since many nodes near the shortest path participate in packet forwarding. This additional overhead caused by moving node can cause congestion in the network.

Author 1: Kohei Arai

Keywords: Blockchain; Ad-hoc network; agent and initiate nodes; the number of hops: connectivity; routing protocol; multi-hop routing; packet forwarding

Paper 11: Recommendation System on Travel Destination based on Geotagged Data

Abstract: Tourism research has benefitted from the worldwide spread and development of social networking services. People nowadays are more likely to rely on internet resources to plan their vacations. Thus, travel recommendation systems are designed to sift through the mammoth amount of data and identify the ideal travel destinations for the users. Moreover, it is shown that the increasing availability and popularity of geotagged data significantly impacts the destination decision. However, most current research concentrates on reviews and textual information to develop the recommendation model. Therefore, the proposed travel recommendation model examines the collective behaviour and connections between users based on geotagged data to provide personalized suggestions for individuals. The model was developed using the user-based collaborative filtering technique. The matrix factorization model was selected as the collaborative filtering technique to compute user similarities due to its adaptability in dealing with sparse rating matrices. The recommendation model generates prediction values to recommend the most appropriate locations. Finally, the model performance of the proposed model was assessed against the popularity and random models using the test design established using Mean Average Precision (MAP), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). The findings indicated that the proposed matrix factorization model has an average MAP of 0.83, with RMSE and MAE values being 1.36 and 1.24, respectively. The proposed model got significantly higher MAP values and the lowest RMSE and MAE values compared to the two baseline models. The comparison shows that the proposed model is effective in providing personalized suggestions to users based on their past visits.

Author 1: Clarice Wong Sheau Harn

Author 2: Mafas Raheem

Keywords: Geotagged data; travel recommendation system; travel recommender; collaborative filtering; matrix factorization

Paper 12: Input Value Chain Affect Vietnamese Rice Yield: An Analytical Model Based on a Machine Learning Algorithm

Abstract: Input value chains greatly affect rice yield, however previous related studies were mainly based on empirical survey and simple statistics, which lacked generality and flexibility. The article presents a new method to predict the influence of input value chain on rice yield in Vietnam based on a machine learning algorithm. Input value chain data is collected through field surveys in rice-growing households. We build a predictive model based on the neural network and swarm intelligence optimization algorithm. The prediction results show that our proposed method has an accuracy of 96%, higher than other traditional methods. This is the basis for management levels to have orientation on the input supply value chain for Vietnamese rice, contributing to the development of the Vietnamese rice brand in the world market.

Author 1: Thi Thanh Nga Nguyen

Author 2: NianSong Tu

Author 3: Thai Thuy Lam Ha

Keywords: Value chains; Vietnamese rice; machine learning; neural network

Paper 13: Cybersecurity in Healthcare: A Review of Recent Attacks and Mitigation Strategies

Abstract: Cyberattacks on several businesses, including those in the healthcare, finance, and industrial sectors, have significantly increased in recent years. Due to inadequate security measures, antiquated practices, and sensitive data, including usernames, passwords, and medical records, the healthcare sector has emerged as a top target for cybercriminals. Cybersecurity has not gotten enough attention in the healthcare sector, despite being crucial for patient safety and a hospital's reputation. In order to prevent data breaches that could jeopardize the privacy of patients' information, hospitals must deploy the proper IT security measures. This research article reviews many scholarly publications that look at ransomware attacks and other cyberattacks on hospitals between 2014 and 2020. The report summarizes the most recent defensive measures put forth in scholarly works that can be used in the healthcare industry. Additionally, the report provides a general review of the effects of cyberattacks and the steps hospitals have taken to manage and recover from these disasters. The study shows that cyberattacks on hospitals have serious repercussions and emphasizes the significance of giving cybersecurity a priority in the healthcare sector. To combat cyberattacks, hospitals must have clear policies and backup plans, constantly upgrade their systems, and instruct employees on how to spot and handle online threats. The article comes to the conclusion that putting in place suitable cybersecurity safeguards can reduce the harm brought on by system failures, reputational damage, and other associated problems.

Author 1: Elham Abdullah Al-Qarni

Keywords: Cybersecurity; healthcare industry; malware; ransomware; DoS; DDoS

Paper 14: Deep Feature Detection Approach for COVID-19 Classification based on X-ray Images

Abstract: The novel human Corona disease (COVID-19) is a pulmonary sickness brought on by an extraordinarily outrageous respiratory condition crown 2. (SARS -CoV-2). Chest radiography imaging has a significant role in the screening, early diagnosis, and follow-up of the suspected individuals due to the effects of COVID-19 on pneumonic-sensitive tissue. It also has a severe impact on the economy as a whole. If positive patients are identified early, the spread of the pandemic illness can be slowed. To determine whether people are at risk for illnesses, a COVID-19 infection prediction is critical. This paper categorizes chest CT samples of COVID-19 affected patients. The two-stage proposed deep learning technique produces spatial function from images, so it is a very expeditious manner for image category hassle. Extensive experiments are drawn by considering the benchmark chest-Computed Tomography (chest-CT) image datasets. Comparative evaluation reveals that our proposed method outperforms amongst other 20 different existing pre-trained models. The test outcomes constitute that our proposed model achieved the best rating of 97.6%, 0.964, 0.964, and 0.982 concerning the accuracy, precision, recall, specificity, and F1-score, respectively.

Author 1: Ayman Noor

Author 2: Priyadarshini Pattanaik

Author 3: Mohammed Zubair Khan

Author 4: Waseem Alromema

Author 5: Talal H. Noor

Keywords: COVID-19; coronavirus; deep learning; classification; chest X-ray images; DenseNet-121; XG-Boost classifier; EfficientNet-B0

Paper 15: Data Sharing using PDPA-Compliant Blockchain Architecture in Malaysia

Abstract: Data privacy is undoubtedly the biggest concern for the modern society. Data privacy is also becoming a key policy in data protection regulations. Organizations assemble massive amount of personal data of the users for monetary and political purposes. These data can be sold for commercial purpose without the prior knowledge or permission from the respective data owners. This can be mitigated by having blockchain to provide a much-needed transparency. However, blockchain’s own transparency becomes its own disadvantage when data owners want to be completely anonymous. Blockchain’s transparent nature will be conflicting with non-linkability. Since the data in blockchain is publicly viewable, any personal data or private transactions being processed through blockchain will be exposed to every node in the network. Hence, blockchain implementations also must comply with privacy acts such as Personal Data Protection Act (PDPA) to have privacy by design and by default. Hence, this paper proposes a PDPA-compliant blockchain architecture for data trading that provides complete control of data and anonymity to the users. A prototype is created using various tools to implement the proposed architecture. This study presents anonymous data sharing for users, data access, data delete features to verify the correctness of the proposed architecture.

Author 1: Hasventhran Baskaran

Author 2: Salman Yussof

Author 3: Asmidar Abu Bakar

Author 4: Fiza Abdul Rahim

Keywords: Blockchain; smart contract; legitimacy; access protocol; retention protocol; data processor; data subject; data user; blockchain regulations

Paper 16: Assessing the Impact and Effectiveness of Cybersecurity Measures in e-Learning on Students and Educators: A Case Study

Abstract: As e-learning has become increasingly prevalent, cyber security has become a major concern. e-Learning platforms collect and store large amounts of sensitive information, such as personal data and financial information, making them attractive targets for cybercriminals. To address these challenges and concerns, e-learning platforms must implement a comprehensive cyber security strategy that includes strong access controls, data encryption, regular software updates, and student training to help them identify and prevent insider threats. This research aims at investigating and determine how satisfied students are with e-learning security and privacy, as well as whether these concerns affect the overall standard of education. A sample study is presented to assess both the impact of the security framework on students' academic achievements and the student’s satisfaction with the security countermeasures in an e-learning system. Statistical analysis showed that the use of security and cyber security countermeasures had a significant effect on the frequent use and participation of students in the contents of the system. Furthermore, encouraging feedback and communication from students about their e-learning experience to share their concerns, questions, and suggestions can help in addressing any security issues or concerns, as well as increasing students’ participation in the e-learning content.

Author 1: Alaa Saeb Al-Sherideh

Author 2: Khaled Maabreh

Author 3: Majdi Maabreh

Author 4: Mohammad Rasmi Al Mousa

Author 5: Mahmoud Asassfeh

Keywords: e-learning; security; cyber security; privacy; countermeasure; Moodle; education

Paper 17: Things of Interest Recommendation with Multidimensional Context Embedding in the Internet of Things

Abstract: The emerging Internet of Things (IoT) makes users and things closely related together, and the interactions between users and things generate massive context data, where the preference information in time, space, and textual content is embedded. Traditional recommendation methods (e.g., movie, music, and location recommendations) are based on static intrinsic context information, which lacks consideration regarding real-time content and spatiotemporal features, failing to adapt to the personalized recommendation in IoT. Therefore, to meet users’ interests and needs in IoT, a novel effective and efficient recommendation method is urgently needed. The paper focuses on mining users’ things of interest in IoT via leveraging multidimensional context embedding. Specifically, to address the challenge from massive context data embedding different user preference information, the paper employs Convolutional Neural Networks (CNN) to mine the intrinsic content information of things and learn their represent. To solve the real-time recommendation problem, the paper proposes a real-time multimodal model embedded into location, time, and some instant content information to track the features of users and things. Furthermore, the paper proposes a matrix factorization-based framework using the regularization method to fuse real-time context embedding and intrinsic information embedding. The experimental results demonstrate the proposed method tailored to IoT is adaptable and flexible, and able to capture user personalized preference effectively.

Author 1: Shuhua Li

Author 2: Jingmin An

Keywords: Internet of things; things of interest; multidimensional context embedding; intrinsic information; instant information; matrix factorization

Paper 18: Reinforcement Learning-based Aspect Term Extraction using Dilated Convolutions and Differential Equation Initialization

Abstract: Aspect term extraction is a crucial subtask in aspect-based sentiment analysis that aims to discover the aspect terms presented in a text. In this paper, a method for ATE is proposed that employs dilated convolution layers to extract feature vectors in parallel, which are then concatenated for classification downstream. Reinforcement learning is used to save the ATE model from imbalance classification, in which the training procedure is posed as a sequential decision-making process. The samples are the states; the network, and the agent; and the agent gets a more significant reward/penalty for correct/incorrect classification of the minority class compared with the majority class. The training phase, which typically employs gradient-based approaches, including back-propagation for the learning process, is tackled. Thus, it suffers from some drawbacks, including sensitivity to initialization. A novel differential equation (DE) approach that uses a clustering-based mutation operator to initiate the BP process is presented. Here, a winning cluster is identified for the current DE population, and a new updating strategy is used to generate candidate solutions. The BERT model is employed as word embedding, which can be included in a downstream task and fine-tuned as a model, while BERT can capture various linguistic properties. The proposed method is evaluated on two English datasets (Restaurant and Laptop) and has achieved outstanding results, surpassing other deep models (Restaurant: Precision 85.44%, F1-score 87.35%; Laptop: Precision 80.88%, F1-score 80.78%).

Author 1: Yuyu Xiong

Author 2: Mariani Md Nor

Author 3: Ye Li

Author 4: Hongxiang Guo

Author 5: Li Dai

Keywords: Aspect term extraction; sentiment analysis; differential evolution; reinforcement learning; BERT

Paper 19: Research on Library Face Book Return Model Based on Hybrid PCA and Kernel Function

Abstract: With the improvement in the quality of university education in China, the behavior of college students and school teachers to borrow and return books in the library is becoming more and more frequent. In peak periods of returning books, managers cannot even assist in returning books in time. Therefore, this research uses kernel function, multi-dimensional principal component analysis method, and multi-dimensional linear discriminant analysis method to construct a new face recognition algorithm for the automatic return of books in the university library. The test results show that the XT_2D_PL algorithm designed in this study has a face recognition rate of 96.8%. When the number of face samples of each type in the test sample set is 11, and when the number of feature dimensions is 14, the recognition rate of 96.3% reaches the highest level. However, if the sample to be processed is 500 pictures, the calculation speed is 1.072ms/per photo, higher than most comparison algorithms. The proposed face recognition algorithm has high recognition accuracy on the library face data; the calculation speed meets the needs of practical applications, and has certain practical promotion potential.

Author 1: Jianwen Shi

Keywords: Kernel function; multidimensional principal component analysis; face recognition; intelligent book return

Paper 20: A Height Accuracy Study Based on RTK and PPK Methods Outside the Standard Working Range

Abstract: The aim of this paper is to study and analyze the Jack Up Vessel (JUV) foundation height accuracy, with the objective of the precise installation of an Offshore Wind Farm (OWF), based on Real Time Kinematic (RTK) and Post-Processing Kinematic (PPK) modes applied on short and long baselines length. The offshore wind farm project is located far from the coastline, not always in the standard working range of RTK. The standard allowed vertical installation tolerance for foundations is less than 10 cm. Taking into account all error sources, deformation of the vessel, motion, lever arms that impact the height measurement of the foundation, it is required that RTK and PPK perform within an accuracy less than 5 cm. In this work, all measures will be evaluated according to the tolerance specification of ±2.5 cm. The survey GNSS tests executed during the project on board of a JUV should be able to provide answers to the following questions: Despite the critical environment, does RTK method allow reaching the theoretical specifications? Does PPK improve accuracy compared to the RTK solution? What is the influence of the baseline length? How much of the time the results fall within the range tolerance? What is the ideal logging period in which accurate and reliable results can be obtained? What is the influence of the hardware and software variants used in testing process on the results accuracy? Based on the test results and analysis a clear description of the influence of different parameters in the OWF precise height measurement in challenging environment will be exposed.

Author 1: Mohamed Jemai

Author 2: Mohamed Anis Loghmari

Author 3: Mohamed Saber Naceur

Keywords: Real time kinematic; post-processing kinematic; high-precision positioning

Paper 21: Analyze Transmission Data from a Multi-Node Patient's Respiratory FMCW Radar to the Internet of Things

Abstract: This is the development of a system that has been made, FMCW radar for human or patient breathing which will then determine the type of disease or disorder in the patient just by looking at the type of breathing. This research uses data from FMCW Radar for human or patient breathing, which is then converted to data that can be read in real-time by the public, doctors, or medical teams through a web server; the web server used is iotmedis.brin.go.id. The novelty of this study is that various types of respiratory data are taken from various points so that it will cause new analysis, namely the process of transmitting data on server traffic or uplink and downlink processes. Specific data and research novelty is how Multi patient respiratory data from OmnipreSense or FMCW Radar can be processed by a microprocessor using MQTT, and multi-patient data can be displayed on the server in real-time.

Author 1: Rizky Rahmatullah

Author 2: Puput Dani Prasetyo Adi

Author 3: Suisbiyanto Prasetya

Author 4: Arief Budi Santiko

Author 5: Yuyu Wahyu

Author 6: B.Berlian Surya Wicaksana

Author 7: Stevry Yushady CH Bissa

Author 8: Riyani Jana Yanti

Author 9: Aloysius Adya Pramudita

Keywords: FMCW Radar data; realtime monitoring; internet of things; transmission data; multi node

Paper 22: Performance Analysis of Prophet Routing Protocol in Delay Tolerant Network by using Machine Learning Models

Abstract: Delay-Tolerant Networking (DTN) or Disruptive-Tolerant Networking comes under the category of networks that works without infrastructure wireless networks. DTN is one type of computer network that provides solutions for several applications. Delay tolerant network communications are networks that are accomplished by storing packets briefly in intermediate nodes till a certain time an end-to-end route is been re-setup or regenerated. This leads to thought as Delay Tolerant Networks. The paper presents the developed models using Artificial Neural Networks (ANN) and Convolutional Neural Networks (CNN) for predicting the best alpha, beta, and gamma parameters of Probabilistic Routing Protocol for Intermittently Connected Networks (PROPHET) protocol for delay tolerant networks. The first data set is generated using ONE simulator, and the generated data is analyzed using python panda’s module. From the above dataset, 80% was used for training and the remaining 20% each has been used for testing and validation. The models were developed and tested using the r2 score for both models to predict alpha, beta, and gamma parameters. Based on the predicted parameters extensive experiments were done and it was found that the ANN model is better than the CNN model. The ANN model can predict optimum alpha, beta, and gamma whereas CNN Model failed to produce accurate prediction.

Author 1: Bonu Satish Kumar

Author 2: Sailaja Vishnubhatla

Author 3: Chevuru Madhu Babu

Author 4: S. Pallam Shetty

Keywords: DTN; ONE; Prophet; CNN; ANN

Paper 23: DIP-CBML: A New Classification of Thai Dragon Fruit Species from Images

Abstract: The attractiveness of dragon fruit is that it has a strange exterior, beautiful colors, and high nutritional value. In Thailand, there is both import and export of dragon fruit. Each package for export must contain only one species of dragon fruit. From the survey, there are seven species of dragon fruit cultivated in Thailand and only some farmers can identify them on his/her farm. Therefore, this research focuses on the classification of Thai dragon fruit from laboratory images and outdoor images; which is different from the previous works which studied only laboratory images. This method was named DIP-CBML that stands for digital image processing with content-based and machine learning. The method consists of image type identification, pre-processing, red and yellow classification, image background removal, and six classes of red species classification. From the results, DIP-CBML can work with both datasets. It gave 100%, 100% and 95.53% accuracy for the image type identification, red and yellow classification, and the classification of six red species respectively. Hopefully, this research will lead to the innovation for the pre-harvest classification of Thai dragon fruit cultivars, applied to industrial applications, and robot harvesting. In the future, may add value to the yield of Thai dragon fruit cultivation.

Author 1: Naruwan Yusamran

Author 2: Nualsawat Hiransakolwong

Keywords: Dragon fruit; classification model; outdoor dataset; image pre-processing; segmentation

Paper 24: Math-VR: Mathematics Serious Game for Madrasah Students using Combination of Virtual Reality and Ambient Intelligence

Abstract: The challenge to increasing understanding of mathematics lessons for students in madrasah schools makes the learning process require the support of adaptive alternative learning media. In this study, we propose a serious game-based learning media supported by virtual reality and ambient intelligence technology to equip students with adaptive responses to subject matter scenarios. Ambient intelligence works based on recommendations generated by the Multi-Criteria Recommender System (MCRS). In calculating a similarity between users and reference data, MCRS uses cosine-based similarity calculations, and average similarity is used for ranking. We developed this learning media experiment called Math-VR using the Unity game engine. The experimental test results show that MCRS-based ambient intelligence technology can provide an adaptive response to the choice of geometry subject matter recommendations for students according to their pre-test results. The analysis results show that the recommendation system as part of ambient intelligence has the highest accuracy rate of 0.92 when using 80 reference data.

Author 1: Hani Nurhayati

Author 2: Yunifa Miftachul Arif

Keywords: Mathematics; serious game; virtual reality; ambient intelligence; MCRS

Paper 25: An Adaptive Channel Selection and Graph ResNet Based Algorithm for Motor Imagery Classification

Abstract: In Brain-Computer interface (BCI) applications, achieving accurate control relies heavily on the classification accuracy and efficiency of motor imagery electroencephalogram (EEG) signals. However, factors such as mutual interference between multi-channel signals, inter-individual variability, and noise interference in the channels pose challenges to motor imagery EEG signal classification. To address these problems, this paper proposes an Adaptive Channel Selection algorithm aimed at optimizing classification accuracy and Information Translate Rate (ITR). First, C3, C4, and Cz are selected as key channels based on neurophysiological evidence and extensive experimental studies. Next, the channel selection is fine-tuned using spatial location and absolute Pearson correlation coefficients. By analyzing the relationship between EEG channels and key channels, the most relevant channel combination is determined for each subject, reducing confounding information and improving classification accuracy. To validate the method, the SHU Dataset and the PhysioNet Dataset are used in experiments. The Graph ResNet classification model is employed to extract features from the selected channel combinations using deep learning techniques. Experimental results show that the average classification accuracy is improved by 5.36% and 9.19%, and the Information Translate Rate is improved by 29.24% and 26.75%, respectively, compared to a single channel combination.

Author 1: Yongquan Xia

Author 2: Jianhua Dong

Author 3: Duan Li

Author 4: Keyun Li

Author 5: Jiaofen Nan

Author 6: Ruyun Xu

Keywords: Brain-Computer Interface; motor imagery; channel selection; deep learning; graph convolutional neural network

Paper 26: Security Challenges Facing Blockchain Based-IoV Network: A Systematic Review

Abstract: The Internet of Vehicles (IoV) is an innovative concept aimed at addressing the critical problem of traffic congestion. IoV applications are part of a connected network that collects relevant data from various smart sensors installed in connected vehicles. This information is freely and easily exchanged between vehicles, which leads to improved traffic management and a reduction in traffic accidents. As the IoV technology continues to grow, the amount of data collected will increase, presenting new challenges for data privacy and security. The use of blockchain technology has been proposed as a solution, as its decentralized and distributed architecture has been proven reliable with cryptocurrencies such as Bitcoin. However, studies have shown that blockchain alone may not be sufficient to address privacy and security concerns, and there are currently no tools available to evaluate its performance in an IoV simulation environment. This research aims to provide a comprehensive review of the challenges associated with the implementation of blockchain technology in the IoV context.

Author 1: Hamza El Mazouzi

Author 2: Anass Khannous

Author 3: Khalid Amechnoue

Author 4: Anass Rghioui

Keywords: Internet of vehicles (IoV); traffic congestion; smart sensors; connected vehicles; data privacy; data security; blockchain technology

Paper 27: Using Descriptive Analysis to Find Patterns and Trends: A Case of Car Accidents in Washington D.C.

Abstract: The descriptive analysis could be used to find the trends and patterns in historical data. In this article, descriptive analysis has been used to describe the car accidents in Washington, D.C., between 2009 and 2020. The dataset was downloaded from the District Department of Transportation (DDOT), the department responsible for car accidents in Washington, D.C. Multiple analytics and statistical models have been applied to find the relationships between different variables and the patterns and trends among the data, such as correlation analysis, confidence interval, One-Way-ANOVA, decision tree, and visualizations. The article aims to find the common reasons for accidents and help DDOT find ways to reduce and eliminate accidents in the area. The statistical and analytical tools examine multiple features to find the patterns and trends among the datasets. Four main findings were found after analyzing the data. First, the main reason for most crashes is drunken people, either drivers or pedestrians. The second finding is that the top reason which causes deadly accidents is speed. Also, we have found that most of the accidents are not dangerous. In addition, we found the top ten streets that contain the highest accident number, and we found that they are located on the north side of the town.

Author 1: Zaid M. Altukhi

Author 2: Nasser F. Aljohani

Keywords: Descriptive analysis; trends; patterns; analytics; statistics; car accidents

Paper 28: Towards an Adaptive e-Learning System Based on Deep Learner Profile, Machine Learning Approach, and Reinforcement Learning

Abstract: Now-a-days, the great challenge of adaptive e-learning systems is to recommend an individualized learning scenario according to the specific needs of learners. Therefore, the perfect adaptive e-learning system is the one that is based on a deep learner profile to recommend the most appropriate learning objects for that learner. Yet, the majority of existing adaptive e-learning systems do not give high importance to the adequacy of the real learner profile and its update with the one taken into account in the learning path recommendation. In this paper, we proposed an intelligent adaptive e-learning system, based on machine learning and reinforcement learning. The objectives of this system are the creation of a deep profile of a given learner, via the implementation of K-means and linear regression, and the recommendation of adaptive learning paths according to this deep profile, by implementing the Q-learning algorithm. The proposed system is decomposed into three principal modules, data preprocessing module, learner deep profile creation module, and learning path recommendation module. These three modules interact with each other to provide a personalized adaptation according to the learner's deep profile. The results obtained indicate that taking into account the learner's deep profile improves the quality of learning for learners.

Author 1: Riad Mustapha

Author 2: Gouraguine Soukaina

Author 3: Qbadou Mohammed

Author 4: Aoula Es-Sâadia

Keywords: Adaptive e-learning system; deep learner profile; reinforcement learning; Q-learning; k-means; linear regression; learning path recommendation; learning object

Paper 29: Industrial Practitioner Perspective of Mobile Applications Programming Languages and Systems

Abstract: The growth of mobile application development industry made it crucial for researchers to study the industry practices of choosing mobile applications programming languages, systems, and tools. With the increased attention of cross-platform mobile applications development from both researchers and industry, this paper aims at answering the question of whether most of the industries are using cross-platform development or native development. The paper collects feedback about industry’s most used mobile development systems. In addition, it provides a map of the different technologies used by mobile applications development companies according to some of the demographics like company size and location. An online questionnaire is carried out to collect the data. The survey targeted both amateur and professional mobile developers. A total of 85 participants participated in answering the survey. Qualitative analysis using descriptive statistics is done on the results of the survey. Although the results show that there is an industrial trend towards using the cross-platform languages, native development is still used by the well-established companies. More than 50% of the participants are found to be aware of the performance issues of the cross-platform development languages and frameworks. Comparison with findings of related work is discussed which raises more research questions and draws out future research in this field.

Author 1: Amira T. Mahmoud

Author 2: Ahmad A. Muhammad

Author 3: Ahmed H. Yousef

Author 4: Hala H.Zayed

Author 5: Walaa Medhat

Author 6: Sahar Selim

Keywords: Android; cross-platform; development; iOS; mobile applications; questionnaire

Paper 30: Feature Selection using Particle Swarm Optimization for Sentiment Analysis of Drug Reviews

Abstract: Feature selection (FS) is an essential classification pre-processing task that eliminates irrelevant, redundant, and noisy features. The primary benefits of performing this task include enhanced model performance, reduced computational expense, and modified “curse of dimensionality”. The goal of performing FS is to find the best feature group that can be used to build an effective pattern recognition model. Drug reviews play a significant role in delivering valuable medical care information, such as the efficacy, side effects, and symptoms of drug use, facilities, drug pricing, and personal drug usage experience to healthcare providers and patients. FS can be used to obtain relevant and valuable information that can produce an optimal subset of features to help obtain accurate results in the classification of drug reviews. The FS approach reduces the number of input variables by eliminating redundant or irrelevant features and narrowing the collection of features to those most significant to the machine learning model. However, the high dimensionality of the feature vector is a major issue that reduces the accuracy of sentiment classification, making it challenging to find the best feature subset. Thus, this article presents a perceptive method to perform FS by gathering information from the potential solutions generated by a particle swarm optimization (PSO) algorithm. This research aimed to apply this algorithm to identify the optimal feature subset of drug reviews to improve the classification accuracy of sentiment analysis. The experimental results showed that PSO provided a better classification performance than a genetic algorithm (GA) and ant colony optimization (ACO) in most datasets. The results showed that PSO demonstrated the highest levels of performance, with an average of 49.3% for precision, 73.6% for recall, 59% for F-score, and 57.2% for accuracy.

Author 1: Afifah Mohd Asri

Author 2: Siti Rohaidah Ahmad

Author 3: Nurhafizah Moziyana Mohd Yusop

Keywords: Sentiment analysis; feature selection; particle swarm optimization; drug reviews

Paper 31: A Comparative Study of Machine Learning Techniques to Predict Types of Breast Cancer Recurrence

Abstract: The prediction of breast cancer recurrence is a crucial problem in cancer research that requires accurate and efficient prediction models. This study aims to compare the performance of different machine learning techniques in predicting types of breast cancer recurrence. In this study, the performance of logistic regression, decision tree, K-Nearest Neighbors, and artificial neural network algorithms was compared on a breast cancer recurrence dataset. The results show that the artificial neural network algorithm outperformed the other algorithms with 91% accuracy, followed by the decision tree (DT) algorithm and K-Nearest Neighbors (kNN) also performed well with accuracies of 90.10% and 88.20%, respectively, while the logistic regression algorithm had the lowest accuracy of 84.60%. The results of this study provide insight into the effectiveness of different machine learning techniques in predicting types of breast cancer recurrence and could guide the development of more accurate prediction models.

Author 1: Meryem Chakkouch

Author 2: Merouane Ertel

Author 3: Aziz Mengad

Author 4: Said Amali

Keywords: Breast cancer; machine learning; recurrence prediction; classification multi-classes; logistic regression; decision tree; K-Nearest Neighbors; artificial neural network

Paper 32: Automated Decision Making ResNet Feed-Forward Neural Network based Methodology for Diabetic Retinopathy Detection

Abstract: The detection of diabetic retinopathy eye disease is a time-consuming and labor-intensive process, that necessitates an ophthalmologist to investigate, assess digital color fundus photographic images of the retina, and discover DR by the existence of lesions linked with the vascular anomalies triggered by the disease. The integration of a single type of sequential image has fewer variations among them, which does not provide more feasibility and sufficient mapping scenarios. This research proposes an automated decision-making ResNet feed-forward neural network methodology approach. The mapping techniques integrated to analyze and map missing connections of retinal arterioles, microaneurysms, venules and dot points of the fovea, cottonwool spots, the macula, the outer line of optic disc computations, and hard exudates and hemorrhages among color and back white images. Missing computations are included in the sequence of vectors, which helps identify DR stages. A total of 5672 sequential and 7231 non-sequential color fundus and black-and-white retinal images were included in the test cases. The 80 and 20 percentage rations of best and poor-quality images were integrated in testing and training and implicated the 10-ford cross-validation technique. The accuracy, sensitivity, and specificity for testing and analysing good-quality images were 98.9%, 98.7%, and 98.3%, and poor-quality images were 94.9%, 93.6%, and 93.2%, respectively.

Author 1: A. Aruna Kumari

Author 2: Avinash Bhagat

Author 3: Santosh Kumar Henge

Author 4: Sanjeev Kumar Mandal

Keywords: Retinal lesion (RL); Fundus Images (FunImg); Microaneurysms (MAs); Principal Component Analysis (PCA); Standard Scaler (StdSca); Feed-Forward Neural Network (FFNN); cross pooling (CxPool)

Paper 33: Proactive Acquisition using Bot on Discord

Abstract: Data deletion increases challenges in cybercrime investigation. To address the problem, proactive forensics for evidentiary collection is acknowledged to help investigators to acquire the potentially needed digital evidence. This study proposes a bot machine to record data from the Discord server in advance, hashing and saving it in proper storage for further forensic analysis. The recording process can be managed to collect activities and their related data (intact, modified, deleted), including text, pictures, videos, and audio. The Discord bot is designed by utilizing the main features of the Discord Social Networks Application Programming Interface (API). This paper examines how this approach is applicable by embedding the bot in a Discord server. Observation showed that the bot records the real-time data as it is always alive on the server, including the deleted or modified messages and their timestamps. All the recorded data is saved locally on the server’s storage in easy-to-read formats, CSV and JSON. The results showed that the bot could conduct the data acquisition for 37 concurrent users with a 2.3% error rate and 97.7% accuracy.

Author 1: Niken Dwi Wahyu Cahyani

Author 2: Daffa Syifa Pratama

Author 3: Nurul Hidayah Ab Rahman

Keywords: Discord; bot; cybercrime; social networks API; digital evidence

Paper 34: The Contribution of Numerical EEG Analysis for the Study and Understanding of Addictions with Substances

Abstract: Computerised electroencephalography (EEG) is one of a wide variety of brain imaging techniques used in addiction medicine. It is a sensitive measure of the effects of addiction on the brain and has been shown to show changes in brain electrical activity during addiction. But, the clinical value of computerised EEG recording in addictions is not yet clearly established. However, several studies argue that this non-invasive technique has an undeniable contribution to the understanding, prediction, diagnosis and monitoring of addictions. The aim of this article is to assess, through a systematic review, the contribution and interest of computerised EEG in the study and understanding of substance abuse by describing the different electrical activities that underlie it across the main frequency ranges: delta, theta, alpha, beta and gamma. We have been conducting a systematic review according to the recommendations of Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) and the Cochrane Group. We included 25 studies with a total of 1897 cases of addiction and 1504 controls. The studies dealt with addictions related to 05 licit and illicit psychoactive substances (alcohol, nicotine, cannabis, heroin and cocaine). The group of addicted patients showed significantly different brain electrical characteristics from the group of controls in the different EEG rhythms, whether during acute substance intoxication, abuse, withdrawal, abstinence, relapse, progression or response to treatment. The majority of studies have used EEG for diagnostic, predictive, monitoring purposes and also to discover electro-physiological markers of certain addictions.

Author 1: Aziz Mengad

Author 2: Jamal Dirkaoui

Author 3: Merouane Ertel

Author 4: Meryem Chakkouch

Author 5: Fatima Elomari

Keywords: Electroencephalography (EEG); quantitative electroencephalography; drug addiction; spectral analysis; coherence analysis

Paper 35: The Usability of Digital Game-based Learning for Low Carbon Awareness: Heuristic Evaluation

Abstract: Digital Game-based Learning (DGBL) that attracts many practitioners to engage students in promoting low carbon awareness has been understudied. The evaluation phase plays a crucial part in determining the usability of the learning material. This study aims to identify the usability of DGBL which consists of four components: game usability (GU), mobility (MO), playability (P), and learning contents (LC) from the perspective of targeted end-user using heuristic evaluation. This study will also provide recommendations to help improve the quality of any related DGBL for novice designers or practitioners. A prototype of DGBL was developed which aims to promote low carbon awareness by learning about fuel cell. The study was designed in two phases, which are (1) developing the heuristic evaluation instrument validated by experts and (2) playtesting to identify the issues of usability in DGBL via heuristic evaluation by targeted end-users, which are forty-six selected students aged fourteen years old. Hence, it shows that the DGBL prototype developed for fuel cell learning has succeeded in achieving learning objectives while promoting low carbon awareness.

Author 1: Nur Fadhilah Abdul Jalil

Author 2: Umi Azmah Hasran

Author 3: Siti Fadzilah Mat Noor

Author 4: Muhammad Helmi Norman

Keywords: Heuristic evaluation; digital game-based learning; usability; low carbon awareness

Paper 36: Optimization Design of Bridge Inspection Vehicle Boom Structure Based on Improved Genetic Algorithm

Abstract: Excessive self-weight of bridge inspection vehicles increases the safety risk of the inspected bridge structures. In this study, a bridge inspection vehicle arm structure self-weight optimization design model is proposed to improve the efficiency and safety of bridge structure inspection. The model uses a finite element model of the arm structure to generate force data to validate and train a back propagation (BP) neural network-based self-weight prediction model of the arm structure, and uses an improved genetic algorithm to assist the prediction model in searching for the optimal solution. The experimental results show that the maximum stress and maximum deformation of the optimal solution from the optimization model designed in this study are lower than the allowable values of the material, and the total weight of the structure from the optimal solution is the lowest, 4687.5 kg. The computational time of the optimization model designed in this study is lower than all the comparison models. The experimental data show that the optimized model for the self-weight optimization of the bridge inspection vehicle arm structure designed in this study has good optimization effect and has some application potential.

Author 1: Ruihua Xue

Author 2: Shuo Lv

Author 3: Tingqi Qiu

Keywords: Genetic algorithm; Bridge inspection; Structural optimization; Finite element model; BP neural network

Paper 37: A Real-Time Automated Visual Inspection System for Printed Circuit Boards Missing Footprints Detection

Abstract: Visual inspection systems (VIS) are vital for recognizing and assessing parts in mass-produced products at the fabricating lines. In the past, item review was carried out physically, which made finding imperfections repetitive, moderate, and prone to error. VIS may be a strategy to abbreviate preparing times, boost item quality, and increment fabricating competitiveness. For the reason of reviewing lost components on uncovered printed circuit sheets, a visual inspection framework is required. The assessment assignment has become more challenging to accomplish the specified quality due to the more compact and complex surface of structured electronic components. This study proposes a real-time visual inspection system to assess lost impressions on Printed Circuit Boards (PCB). This system is composed of hardware and software frameworks. The main contribution of this study is the proposed software framework. The software framework consists of components region analysis and missing detection using image processing, cross-correlation, and production rules. Experimental results show the viability and achievability of the proposed system for PCB missing component detection.

Author 1: Xiaoda Cao

Keywords: Automated visual inspection; Printed Circuit Boards (PCB); quality control; image processing

Paper 38: PM2.5 Estimation using Machine Learning Models and Satellite Data: A Literature Review

Abstract: Most researchers are beginning to appreciate the use of remote sensing satellites to assess PM2.5 levels and use machine learning algorithms to automate the collection, make sense of remote sensing data, and extract previously unseen data patterns. This study reviews delicate particulate matter (PM2.5) predictions from satellite aerosol optical depth (AOD) and machine learning. Specifically, we review the characteristics and gap-filling methods of satellite-based AOD products, sources and components of PM2.5, observable AOD products, data mining, and the application of machine learning algorithms in publications of the past two years. The study also included functional considerations and recommendations in covariate selection, addressing the spatiotemporal heterogeneity of the PM2.5 -AOD relationship, and the use of cross-validation, to aid in determining the final model. A total of 79 articles were included out of 112 retrieved records consisting of articles published in 2022 totaling 43 articles, as of 2023 (until February) totaling 19 articles, and other years totaling 18 articles. Finally, the latest method works well for monthly PM2.5 estimates, while daily PM2.5 and hourly PM2.5 can also be achieved. This is due to the increased availability and computing power of large datasets and increased awareness of the potential benefits of predictors working together to achieve higher estimation accuracy. Some key findings are also presented in the conclusion section of this article.

Author 1: Mitra Unik

Author 2: Imas Sukaesih Sitanggang

Author 3: Lailan Syaufina

Author 4: I Nengah Surati Jaya

Keywords: AOD; machine learning; PM2.5; remote sensing; pollutant

Paper 39: Developing A Predictive Model for Selecting Academic Track Via GPA by using Classification Algorithms: Saudi Universities as Case Study

Abstract: The main motivation of any educational institution is to provide quality education. Therefore, choosing an academic track can be clearly seen as an obstacle, for students and universities, which in turn led to imposing a mandatory preparatory year program in Saudi Arabia. One of the main objectives of the preparatory year is to help students discover the right academic track. Nevertheless, some students choose the wrong academic track which can be a stumbling block that may prevent their progress. According to the tremendous growth of using information technology, educational data mining technology (EDM) can be applied to discover useful patterns, unlike traditional data analysis methods. Most of the previous research focused on predicting the GPA after the students choose an academic track. On the contrary, our research focuses on using classification algorithms to develop a predictive model for advising students to select academic tracks via prediction of the GPA based on the preparatory year data at Saudi Universities. Then, compare classification algorithms to provide the most accurate prediction. The dataset was extracted from a Saudi university containing preparatory year data for 2363 students. This work was carried out using five classification algorithms: Gradient Boosting(GB), K-Nearest Neighbors (kNN), Logistic Regression (LG), Neural Network(NN) and Random Forest(RF). The results showed the superiority of the Logistic Regression algorithm in terms of accuracy over the other algorithms. Future work could add behavioral characteristics of students and use other algorithms to provide better accuracy.

Author 1: Thamer Althubiti

Author 2: Tarig M. Ahmed

Author 3: Madini O. Alassafi

Keywords: Data mining; educational data mining; classification algorithms; logistic regression; neural networks; gradient boosting; k-nearest neighbors; predicting students’ performance

Paper 40: Combining GAN and LSTM Models for 3D Reconstruction of Lung Tumors from CT Scans

Abstract: Generating a three-dimensional (3D) reconstruction of tumors is an efficient technique for obtaining accurate and highly detailed visualization of the structures of tumors. To create a 3D tumor model, a collection of 2D imaging data is required, including images from CT imaging. Generative adversarial networks (GANs) offer a method to learn helpful representations without annotating the training dataset considerably. The article proposes a technique for creating a 3D model of lung tumors from CT scans using a combination of GAN and LSTM models, with support from ResNet as a feature extractor for the 2D images. The model presented in this article involves three steps, starting with the segmentation of the lung, then the segmentation of the tumor, and concluding with the creation of a 3D reconstruction of the lung tumor. The segmentation of the lung and tumor is conducted utilizing snake optimization and Gustafson–Kessel (GK) method. To prepare the 3D reconstruction component for training, the ResNet model that has been pre-trained is utilized to capture characteristics from 2D lung tumor images. Subsequently, the series of characteristics that have been extracted are fed into a LSTM network to generate compressed features as the final output. Ultimately, the condensed feature is utilized as input for the GAN framework, in which the generator is accountable for generating a sophisticated 3D lung tumor image. Simultaneously, the discriminator evaluates whether the 3D lung tumor image produced by the generator is authentic or synthetic. This model is the initial attempt that utilizes a GAN model as a means for reconstructing 3D lung tumors. The suggested model is evaluated against traditional approaches using the LUNA dataset and standard evaluation metrics. The empirical findings suggest that the suggested approach shows a sufficient level of performance in comparison to other methods that are vying for the same objective.

Author 1: Cong Gu

Author 2: Hongling Gao

Keywords: 3D tumor reconstruction; lung cancer; LSTM; Generative adversarial network; ResNet

Paper 41: Optimized Secure Federated Learning for Event Detection in Big Data using Blockchain Mechanism

Abstract: Currently, cloud storage in blockchain and federated learning technology provides better security among data transmission and file access. But, in some of the cases, security issues arose. So, to avoid security problems and offer better protection in a cloud environment, a novel optimized buffalo-based Homomorphic SHA blockchain (OBHSB). In this model for accessing the cloud storage data with the key matching method, if any of the unauthenticated users are trying to access the file initially, the system checks the key matching parameter. The proposed model was developed to provide better security in big data presented in the cloud environment. However, the parameters in the proposed model were compared with the existing models to make sure better performance was attained through the proposed model. Attack was considered as an event in this research. In the performance analysis, the performance rate of the proposed model was validated. Subsequently, the case study was developed in this research to explain the working procedure of the proposed design; model performs hashing, encryption, decryption, and key matching mechanisms. The results proposed model is observed to have 100% confidentiality rate after attack.

Author 1: K. Prasanna Lakshmi

Author 2: K. Swapnika

Keywords: Blockchain; cloud storage; decryption; encryption; federated learning; hashing; homomorphism

Paper 42: Hate Speech Detection in Social Networks using Machine Learning and Deep Learning Methods

Abstract: Hate speech on social media platforms like Twitter is a growing concern that poses challenges to maintaining a healthy online environment and fostering constructive communication. Effective detection and monitoring of hate speech are crucial for mitigating its adverse impact on individuals and communities. In this paper, we propose a comprehensive approach for hate speech detection on Twitter using both traditional machine learning and deep learning techniques. Our research encompasses a thorough comparison of these techniques to determine their effectiveness in identifying hate speech on Twitter. We construct a robust dataset, gathered from diverse sources and annotated by experts, to ensure the reliability of our models. The dataset consists of tweets labeled as hate speech, offensive language, or neutral, providing a more nuanced representation of online discourse. We evaluate the performance of LSTM, BiLSTM, and CNN models against traditional shallow learning methods to establish a baseline for comparison. Our findings reveal that deep learning techniques outperform shallow learning methods, with BiLSTM emerging as the most accurate model for hate speech detection. The BiLSTM model demonstrates improved sensitivity to context, semantic nuances, and sequential patterns in tweets, making it adept at capturing the intricate nature of hate speech. Furthermore, we explore the integration of word embeddings, such as Word2Vec and GloVe, to enhance the performance of our models. The incorporation of these embeddings significantly improves the models' ability to discern between hate speech and other forms of online communication. This paper presents a comprehensive analysis of various machine learning methods for hate speech detection on Twitter, ultimately demonstrating the superiority of deep learning techniques, particularly BiLSTM, in addressing this critical issue. Our findings pave the way for further research into advanced methods of tackling hate speech and facilitating healthier online interactions.

Author 1: Aigerim Toktarova

Author 2: Dariga Syrlybay

Author 3: Bayan Myrzakhmetova

Author 4: Gulzat Anuarbekova

Author 5: Gulbarshin Rakhimbayeva

Author 6: Balkiya Zhylanbaeva

Author 7: Nabat Suieuova

Author 8: Mukhtar Kerimbekov

Keywords: Machine learning; deep learning; hate speech; social network; classification

Paper 43: New Arabic Root Extraction Algorithm

Abstract: This research presents a new algorithm for Arabic root extraction, which aims to improve the accuracy of Arabic Natural Language Processing Algorithms by addressing the weaknesses and errors of existing algorithms. The proposed algorithm utilizes a database, that includes a collection of roots, patterns, and affixes, to generate potential derivation roots of a word without eliminating affixes initially. By matching the derived word with patterns to identify potential roots, the proposed algorithm avoids the inaccuracies caused by eliminating affixes based on expectation methods. The study conducted a comparison of the proposed algorithm with three commonly used Arabic root extraction algorithms. The evaluation process is performed on three corpora. Results showed that the proposed algorithm achieved an average accuracy rate of 96%, which is significantly higher than the others.

Author 1: Nisrean Jaber Thalji

Author 2: Emran Aljarrah

Author 3: Roqia Rateb

Author 4: Amaal Rateb Mohammad Al-Shorman

Keywords: Natural language processing; Arabic root extraction algorithm; Arabic applications; Arabic morphology; Text mining

Paper 44: An Evolutive Knowledge Base for “AskBot” Toward Inclusive and Smart Learning-based NLP Techniques

Abstract: Artificial Intelligence chatbots have shown a growing interest in different domains including e-learning. They support learners by answering their repetitive and massive questions. In this paper, we develop a smart learning architecture for an inclusive chatbot handling both text and voice messages. Thus, disabled learners can easily use it. We automatically extract, preprocess, vectorize, and construct AskBot's Knowledge Base. The present work evaluates various vectorization techniques with similarity measures to answer learners' questions. The proposed architecture handles both Wh-Questions starting with Wh words and Non-Wh-Questions, beginning with unpredictable words. Regarding Wh-Questions, we develop a neural network model to classify intents. Our results show that the model's accuracy and the F1-Score are equal to 99,5%, and 97% respectively. With a similarity score of 0.6, our findings indicate that TF-IDF has performed well, correctly answering 90% of the tested Wh-Questions. Concerning No-Wh Questions, soft cosine measure, and fasttext successfully answered 72% of Non-Wh-Question.

Author 1: Khadija El Azhari

Author 2: Imane Hilal

Author 3: Najima Daoudi

Author 4: Rachida Ajhoun

Author 5: Ikram Belgas

Keywords: Knowledge base; KB; artificial intelligence; AI; chatbot; e-learning; cosine similarity; soft cosine similarity; TF-IDF; FastText; neural network

Paper 45: Knowledge Management Model for the Generation of Innovative Capacities in Organizations that Provide Services

Abstract: The research was oriented to the development of a knowledge management model for the generation of innovative capacities in the organizations that provide services. A systematic review of articles published in the Scopus, IEEE Explore and Google Scholar databases was carried out, where 67 articles and 24 models were selected, which were subsequently analyzed based on their theoretical foundation, strategies used for the generation and dissemination of knowledge, incorporation of the organizational culture and the use of Information and Communication Technology (ICT) in the generation and dissemination of knowledge. The proposed model, unlike the models evaluated, is oriented towards generating added value with a new strategic approach structured in the knowledge management and organizational memory macro-processes, which in turn are divided into 29 and 11 macro-activities respectively, which incorporate the organizational culture and allows guiding the organization to improve its functions through the incorporation of innovation and use of ICT in all processes of the organization and in each stage of the generation and management of knowledge; establishing the essential parameters for the generation of innovative capacities, generation of knowledge, intellectual capital and transfer of information to knowledge, which can be used within the organization. The proposed model, unlike the models evaluated, is aimed at directly strengthening interpersonal relationships between members of the organization and between them and their clients. In the same way, it incorporates a maturity model made up of five levels to measure the state in which the organization is in relation to knowledge management.

Author 1: Cristhian Ronceros

Author 2: José Medina

Author 3: Pedro León

Author 4: Alfredo Mendieta

Author 5: José Fernández

Author 6: Yuselys Martinez

Keywords: Component; model; knowledge management; intellectual capital; information and communication technologies

Paper 46: Recognition of Lung Nodules in Computerized Tomography Lung Images using a Hybrid Method with Class Imbalance Reduction

Abstract: Lung cancer is among the deadly diseases affecting millions globally every year. Physicians' and radiologists' manual detection of lung nodules has low efficiency due to the variety of shapes and nodule locations. The paper aims to recognize the lung nodules in computerized tomography (CT) lung images utilizing a hybrid method to reduce the problem space at every step. First, the suggested method uses the fast and robust fuzzy c-means clustering method (FRFCM) algorithm to segment CT images and extract two lungs, followed by a convolutional neural network (CNN) to identify the sick lung for use in the next step. Then, the adaptive thresholding method detects the suspected regions of interest (ROIs) among all available objects in the sick lung. Next, some statistical features are selected from every ROI, and then a restricted Boltzmann machine (RBM) is considered a feature extractor that extracts rich features among the selected features. After that, an artificial neural network (ANN) is employed to classify ROIs and determine whether the ROI includes nodules or non-nodules. Finally, cancerous ROIs are localized by the Otsu thresholding algorithm. Naturally, sick ROIs are more than healthy ones, leading to a class imbalance that substantially decreases ANN ability. To solve this problem, a reinforcement learning (RL)-based algorithm is used, in which the states are sampled. The agent receives a larger reward/penalty for correct/incorrect classification of the examples related to the minority class. The proposed model is compared with state-of-the-art methods on the lung image database consortium image collection (LIDC-IDRI) dataset and standard performance metrics. The results of the experiments demonstrate that the proposed model outperforms its rivals.

Author 1: Yingqiang Wang

Author 2: Honggang Wang

Author 3: Erqiang Dong

Keywords: Lung cancer; artificial neural network; fuzzy c-means clustering method; reinforcement learning; restricted boltzmann machine

Paper 47: A Theoretical Framework for Creating Folk Dance Motion Templates using Motion Capture

Abstract: Folk dance (FD) is a type of traditional dance that has been handed down through a culture or group from generation to generation. It is crucial to preserve and safeguard this type of cultural legacy since it can reflect the history and identity of particular nations. However, due to ineffective preservation and conservation techniques, the survival of FDs is being negatively impacted more and more. Its extinction may be caused by ignorance about and disregard for preservation and conservation efforts. The most efficient method for digitizing intangible cultural property, including FDs, is motion capture (MoCap). MoCap enables the conversion of real-time movement into digital performance to produce motion templates. This paper aims to provide suggestions and guidelines in conducting research to generate motion templates of FDs. Several key approaches are presented and discussed in detail, including acquaintance meetings, procedures and approval, interviews and experiments, and the framework. The proposed framework includes models for MoCap, skeleton generation, skeleton refinement, and evaluation. By implementing the proposed framework, the motion templates for FDs can be created. The generated motion templates will preserve and conserve FDs and guarantee their originality and authenticity.

Author 1: Amir Irfan Mazian

Author 2: Wan Rizhan

Author 3: Normala Rahim

Author 4: Azrul Amri Jamal

Author 5: Ismahafezi Ismail

Author 6: Syed Abdullah Fadzli

Keywords: Motion capture; folk dance; motion template

Paper 48: Development of a New Lightweight Encryption Algorithm

Abstract: Due to the growing need to use devices with low hardware resources in everyday life, the likelihood of their susceptibility to various cyber-attacks increases. In this regard, one of the methods to ensure the security of information circulating in these devices is encryption. For devices with small hardware resources, the most applicable is low-resource (lightweight) cryptography. This article introduces a new lightweight encryption algorithm, ISL-LWS (Information Security Laboratory – lightweight system), designed to protect data on resource-constrained devices. The encryption algorithm is implemented in the C++ programming language. The paper presents the statistical properties of ciphertexts obtained using the developed algorithm. For the experimental testing for statistical security, the sets of statistical tests by NIST and D. Knuth were used. Separately, the ISL-LWS algorithm was tested for avalanche effect properties. The obtained results of statistical tests were compared with the Present and Speck modern lightweight algorithms. The study and comparative analysis of the speed of encryption and key generation of the three algorithms were carried out on the Arduino Uno R3 board.

Author 1: Ardabek Khompysh

Author 2: Nursulu Kapalova

Author 3: Oleg Lizunov

Author 4: Dilmukhanbet Dyusenbayev

Author 5: Kairat Sakan

Keywords: Lightweight block cipher; S-box; linear transformation; avalanche effect; IoT devices; RFID tags; null hypothesis; NIST tests; D. knuth tests

Paper 49: An Investigation of Asthma Experiences in Arabic Communities Through Twitter Discourse

Abstract: Artificial intelligence technologies can effectively analyze the public opinions from social-media platforms like twitter. This study aims to employ the AI technology and big data to explore and discuss the common issues of asthma that patients share on Twitter platform in Arabic communities. The data was acquired using the Twitter API version 2. Latent Dirichlet Allocation was used for grouping data into two clusters which provide information and tips about the treatment and prevention of asthma and personal experiences with asthma, including symptoms, diagnosis, and the negative impact of asthma on the quality of life. Sentiment analysis and data frequency distribution techniques were used to analyze the data in both clusters. The data analysis of first indicated that individuals are interested in learning about different ways to treat asthma and potentially finding a permanent solution. The data analysis of second cluster indicated the existence of negative sentiments about asthma, which also included religious expressions for improving the condition. The study also discussed the differences in expressions among Arabic communities and other communities.

Author 1: Mohammed Alotaibi

Author 2: Ahmed Omar

Keywords: Asthma; twitter; semantic analysis; LDA; Arab; communities

Paper 50: Predicting Drug Response on Multi-Omics Data Using a Hybrid of Bayesian Ridge Regression with Deep Forest

Abstract: An accurate drug response prediction for each patient is critical in personalized medicine. However, numerous studies that relied on single-omics datasets continue to have limitations. In addition, the curse of dimensionality considers a challenge to drug response prediction. Deep learning has remarkable prediction effectiveness compared to traditional machine learning, but it requires enormous amounts of training data which is a limitation because the nature of most biological data is small-scale. This paper presents an approach that combines Bayesian Ridge Regression with Deep Forest. BRR relies on the Bayesian approach, in which linear model estimation occurs based on probability distributions rather than point estimates. It was utilized to integrate multi-omics, a feature selection that calculates the coefficient as the feature importance. DF reduces the computational cost and hyper-parameter tuning cost. The Cancer Cell Line Encyclopedia CCLE was used as a dataset to integrate the gene expression, copy number variant, and single nucleotide variant. Root Mean Square Error, Pearson Correlation Coefficient, and the coefficient of determination were used as the evaluation metrics. The obtained findings show that the proposed model outperforms Random Forest and Convolutional Neural Network regarding regression performance; it achieved 0.175 for RMSE, 0.842 for PCC, and 0.708 for R2.

Author 1: Talal Almutiri

Author 2: Khalid Alomar

Author 3: Nofe Alganmi

Keywords: Bayesian ridge regression; deep forest; deep learning; drug response prediction; machine learning; multi-omics data

Paper 51: Systematic Analysis on the Effectiveness of Covert Channel Data Transmission

Abstract: A covert channel is a communication channel that allows parties to communicate and transfer data indirectly. Covert channel types are storage, timing, and behavior channels. Covert channels can be used for malicious and benign applications. A covert channel is a mechanism for violating the communication security policy that was not anticipated by the system creator. Recently, covert channels are used to transfer text, video, and audio information between entities. This article, discusses studies related to the development of covert channels as well as the research works that focus on improving the performance/throughput of covert channels. Also, it analyzes the previous studies in terms of publication type, year of publication, article title, article purpose, transferring file format used in covert channel, coding technique, throughput performance, time needed to transfer files, and article limitations.

Author 1: Abdulrahman Alhelal

Author 2: Mohammed Al-Khatib

Keywords: Covert channel; transmission; limitations; file; encoding throughput; performance; time; audio; video; text

Paper 52: Towards Analysis of Biblical Entities and Names using Deep Learning

Abstract: Scholars from various fields have studied the translations of the Bible in different languages to understand the changes that have occurred over time. Taking into account recent advances in deep learning, there is an opportunity to improve the understanding of these texts and conduct analyses that were previously unattainable. This study used deep learning techniques of NLP to analyze the distribution and appearance of names in the Polish, Croatian, and English translations of the Gospel of Mark. Within the scope of social network analysis (SNA), various centrality metrics were used to determine the importance of different entities (names) within the gospel. Degree Centrality, Closeness Centrality, and Betweenness Centrality were leveraged, given their capacity to provide unique insights into the network structure. The findings of this study demonstrate that deep learning could help uncover interesting connections between individuals who may have initially been considered less important. It also highlighted the critical role of onomastic sciences and the philosophy of language in analyzing the richness and importance of human and other proper names in biblical texts. Further research should be conducted to produce more relevant language resources, improve parallel multilingual corpora and annotated data sets for the major languages of the Bible, and develop an accurate end-to-end deep neural model that facilitates joint entity recognition and resolution.

Author 1: Mikolaj Martinjak

Author 2: Davor Lauc

Author 3: Ines Skelac

Keywords: Bible; deep learning; gospel of mark; natural language processing; social network analysis

Paper 53: Deadline-aware Task Scheduling for Cloud Computing using Firefly Optimization Algorithm

Abstract: Task scheduling poses a major challenge for cloud computing environments. Task scheduling ensures cost-effective task execution and improved resource utilization. It is classified as a NP-hard problem due to its nondeterministic polynomial time nature. This characteristic motivates researchers to employ meta-heuristic algorithms. The number of cloud users and computing capabilities is leading to increased concerns about energy consumption in cloud data centers. In order to leverage cloud resources in the most energy-efficient manner while delivering real-time services to users, a viable cloud task scheduling solution is necessary. This study proposes a new deadline-aware task scheduling algorithm for cloud environments based on the Firefly Optimization Algorithm (FOA). The suggested scheduling algorithm achieves a higher level of efficiency in multiple parameters, including execution time, waiting time, resource utilization, the percentage of missed tasks, power consumption, and makespan. According to simulation results, the proposed algorithm is more effective and superior to the CSO algorithm under HP2CN and NASA workload archives.

Author 1: BAI Ya-meng

Author 2: WANG Yang

Author 3: WU Shen-shen

Keywords: Cloud computing; energy efficiency; task scheduling; firefly algorithm

Paper 54: A Method for Network Intrusion Detection Based on GAN-CNN-BiLSTM

Abstract: As network attacks are more and more frequent and network security is more and more serious, it is important to detect network intrusion accurately and efficiently. With the continuous development of deep learning, a lot of research achievements are applied to intrusion detection. Deep learning is more accurate than machine learning, but in the face of a large amount of data learning, the performance will be degraded due to data imbalance. In view of the serious imbalance of network traffic data sets at present, this paper proposes to process data expansion with GAN to solve data imbalance and detect network intrusion in combination with CNN and BiLSTM. In order to verify the efficiency of the model, the CIC-IDS 2017 data set is used for evaluation, and the model is compared with machine learning methods such as Random Forest and Decision Tree. The experiment shows that the performance of this model is significantly improved over other traditional models, and the GAN-CNN-BiLSTM model can improve the efficiency of intrusion detection, and its overall accuracy is improved compared with SVM, DBN, CNN, BiLSTM and other models.

Author 1: Shuangyuan Li

Author 2: Qichang Li

Author 3: Mengfan Li

Keywords: Intrusion detection; GAN; CNN; BiLSTM

Paper 55: A Consumer Product of Wi-Fi Tracker System using RSSI-based Distance for Indoor Crowd Monitoring

Abstract: This study aims to design and develop Wi-Fi tracker system that utilizes RSSI-based distance parameters for crowd-monitoring applications in indoor settings. The system consists of three main components, namely 1) an embedded node that runs on Raspberry-pi Zero W, 2) a real-time localization algorithm, and 3) a server system with an online dashboard. The embedded node scans and collects relevant information from Wi-Fi-connected smartphones, such as MAC data, RSSI, timestamps, etc. These data are then transmitted to the server system, where the localization algorithm passively determines the location of devices as long as Wi-Fi is enabled. The mentioned devices are smartphones, tablets, laptops, while the algorithm used is a Non-Linear System with Lavenberg–Marquart and Unscented Kalman Filter (UKF). The server and online dashboard (web-based application) have three functions, including displaying and recording device localization results, setting parameters, and visualizing analyzed data. The node hardware was designed for minimum size and portability, resulting in a consumer electronics product outlook. The system demonstration in this study was conducted to validate its functionality and performance.

Author 1: Syifaul Fuada

Author 2: Trio Adiono

Author 3: Prasetiyo

Author 4: Harthian Widhanto

Author 5: Shorful Islam

Author 6: Tri Chandra Pamungkas

Keywords: Wi-Fi tracker system; RSSI-based distance; crowd monitoring; Unscented Kalman Filter; indoor

Paper 56: A Knowledge Based Framework for Cardiovascular Disease Prediction

Abstract: Cardiovascular disease has become more concern in the hectic and stressful life of modern era. Machine learning techniques are becoming reliable in medical treatment to help the doctors. But the ML algorithms are sensitive to data sets. Hence a Smart Robust Predictive System is almost essential which can work efficiently on all data sets. The study proposes ensembled classifier validating its performance on five different data sets- Cleveland, Hungarian, Long Beach, Statlog and Combined datasets. The developed model deals with missing values and outliers. Synthetic Minority Oversampling Technique (SMOTE) was used to resolve the class imbalance issue. In this study, performance of five individual classifiers – Support Vector Machine Radial (SVM-R), Logistic Regression (LR), Naïve Bayes (NB), Random Forest (RF) and XGBoost, was compared with five ensembled classifiers on five different data sets. On each data set the top three performers were identified and were combined to give ensemble classifiers. Thus, in all total 25 experimentation were done. The results have shown that out of all classifiers implemented, the proposed system outperforms on all the data sets. The performance was validated by 10-fold cross validation The proposed system gives the highest accuracy and sensitivity of 87% and 86% respectively.

Author 1: Abha Marathe

Author 2: Virendra Shete

Author 3: Dhananjay Upasani

Keywords: Machine learning; ensemble classifier; cardiovascular disease; performance metrics; classifier techniques

Paper 57: Unsupervised Bearing Fault Diagnosis via a Multi-Layer Subdomain Adaptation Network

Abstract: Bearings play a crucial role in the functioning of rotating machinery, making it essential to monitor their condition for maintaining system stability and dependability. In recent years, intelligent diagnostic techniques for bearing issues have made significant progress due to advancements in artificial intelligence. These methods rely heavily on data, requiring data collection and labeling to develop the learning model, which is often highly challenging and nearly infeasible in industrial settings. As a result, a domain adaptation-based transfer learning approach has been suggested. This approach aims to minimize the difference between the distribution of accessible data and the unlabeled real-world data, enabling the model trained on public data to function effectively with actual data. In this paper, we introduce a sophisticated subdomain adaptation technique for cross-machine bearing fault diagnosis using vibration, termed multi-layer subdomain adaptation. Verification experiments were conducted, and the findings indicate that the proposed approach offers relatively high accuracy up to 97.47% and excellent transferability. Comparative experiments revealed that the proposed method is a superior technique for bearing fault diagnosis and slightly outperforms other methods (3-5%) in both predictive and noise-ignore capabilities. Comprehensive validation experiments were conducted using the HUST dataset.

Author 1: Nguyen Duc Thuan

Author 2: Nguyen Thi Hue

Author 3: Hoang Si Hong

Keywords: Bearing fault; fault diagnosis; domain adaptation; transfer learning

Paper 58: Mask R-CNN Approach to Real-Time Lane Detection for Autonomous Vehicles

Abstract: The accurate and real-time detection of road lanes is crucial for the safe navigation of autonomous vehicles (AVs). This paper presents a novel approach to lane detection by leveraging the capabilities of the Mask Region-based Convolutional Neural Network (Mask R-CNN) model. Our method adapts Mask R-CNN to specifically address the challenges posed by diverse traffic scenarios and varying environmental conditions. We introduce a robust, efficient, and scalable architecture for lane detection, which segments the lane markings and generates precise boundaries for AVs to follow. We augment the model with a custom dataset, consisting of images collected from different geographical locations, weather conditions, and road types. This comprehensive dataset ensures the model's generalizability and adaptability to real-world conditions. We also introduce a multi-scale feature extraction technique, which improves the model's ability to detect lanes in both near and far fields of view. Our proposed method significantly outperforms existing state-of-the-art techniques in terms of accuracy, processing speed, and adaptability. Extensive experiments were conducted on public datasets and our custom dataset to validate the performance of the proposed method. Results demonstrate that our Mask R-CNN-based approach achieves high precision and recall rates, ensuring reliable lane detection even in complex traffic scenarios. Additionally, our model's real-time processing capabilities make it an ideal solution for implementation in AVs, enabling safer and more efficient navigation on roads.

Author 1: Rustam Abdrakhmanov

Author 2: Madina Elemesova

Author 3: Botagoz Zhussipbek

Author 4: Indira Bainazarova

Author 5: Tursinbay Turymbetov

Author 6: Zhalgas Mendibayev

Keywords: Road; lane; Mask R-CNN; detection; deep learning; autonomous vehicle

Paper 59: Game Theory Approach for Open Innovation Systems Analysis in Duopolistic Market

Abstract: The approach used in this study involves applying the Cournot model, which is initially based on the analysis of product quantities in the market. Building upon the obtained equilibrium, a second analysis is conducted to examine the impact of the open innovation integration rate, utilizing a dynamic model. The obtained results have demonstrated that multiple equilibria are possible, and under certain conditions, competing firms have a stake in carefully analyzing the integration rate of open innovation.

Author 1: Aziz Elmire

Author 2: Aziz Ait Bassou

Author 3: Mustapha Hlyal

Author 4: Jamila El Alami

Keywords: Duopoly; open innovation; closed innovation; Cournot model

Paper 60: Decentralised Access Control Framework using Blockchain: Smart Farming Case

Abstract: The convergence of farming with cutting-edge technologies, like the Internet of Things (IoT), has led to the emergence of a smart farming revolution. IoT facilitates the interconnection of numerous devices across different agricultural ecosystems, enabling automation and ultimately enhancing the efficiency and quality of production. However, the implementation of IoT entails an array of potential risks. The accelerated adoption of IoT in the domain of smart farming has amplified the existing cybersecurity concerns, specifically those pertaining to access control. In extensive IoT environments that require scalability, the conventional centralized access control system is insufficient. Therefore, to address these gaps, we propose a novel decentralized access control framework. The framework applies blockchain technology as the decentralization approach with smart contract application focuses on the application scenario in smart farming to protect and secure IoT devices from unauthorised access by anomalous entities. The proposed framework adopted attribute-based access control (ABAC) and role-based access control (RBAC) to establish access rules and access permissions for IoT. The framework is validated via simulation to determine the price of gas consumption when executing smart contracts to retrieve attributes, roles and access rules between three smart contracts and provide the baseline value for future research references. Thus, this paper offers valuable insight into ongoing research on decentralized access control for IoT security to protect and secure IoT resources in the smart farming environment.

Author 1: Normaizeerah Mohd Noor

Author 2: Noor Afiza Mat Razali

Author 3: Sharifah Nabila S Azli Sham

Author 4: Khairul Khalil Ishak

Author 5: Muslihah Wook

Author 6: Nor Asiakin Hasbullah

Keywords: Access control; role-based access control; attribute-based access control; blockchain technology; internet of things; smart contract; smart farming

Paper 61: Artificial Intelligence System for Detecting the Use of Personal Protective Equipment

Abstract: In recent years, occupational accidents have been increasing, and it has been suggested that this increase is related to poor or no supervision of personal protective equipment (PPE) use. This study proposes developing a system capable of identifying the use of PPEs using artificial intelligence through a neural network called YOLO. The results obtained from the development of the system suggest that automatic recognition of PPEs using artificial intelligence is possible with high precision. The recognition of gloves is the only critical object that can give false positives, but it can be addressed with a redundant system that performs two or more consecutive recognitions. This study also involved the preparation of a custom dataset for training the YOLO neural network. The dataset includes images of workers wearing different types of PPEs, such as helmets, gloves, and safety shoes. The system was trained using this dataset and achieved a precision of 98.13% and a recall of 86.78%. The high precision and recall values indicate that the system can accurately identify the use of PPEs in real-world scenarios, which can help prevent occupational accidents and ensure worker safety.

Author 1: Josue Airton Lopez Cabrejos

Author 2: Avid Roman-Gonzalez

Keywords: Personal protective equipment (PPE); artificial intelligence (AI); YOLO (You Only Look Once); object detection; neural network; custom PPE dataset

Paper 62: The Use of Fuzzy Linear Regression Modeling to Predict High-risk Symptoms of Lung Cancer in Malaysia

Abstract: Lung cancer is the most prevalent cancer in the world, accounting for 12.2% of all newly diagnosed cases in 2020 and has the highest mortality rate due to its late diagnosis and poor symptom detection. Currently, there are 4,319 lung cancer deaths in Malaysia, representing 2.57 percent of all mortality in 2020. The late diagnosis of lung cancer is common, which makes survival more difficult. In Malaysia, however, most cases are detected when the tumors have become too large, or cancer has spread to other body areas that cannot be removed surgically. This is a frequent situation due to the lack of public awareness among Malaysians regarding cancer-related symptoms. Malaysians must be acknowledged the high-risk symptoms of lung cancer to enhance the survival rate and reduce the mortality rate. This study aims to use a fuzzy linear regression model with heights of triangular fuzzy by Tanaka (1982), H-value ranging from 0.0 to 1.0, to predict high-risk symptoms of lung cancer in Malaysia. The secondary data is analyzed using the fuzzy linear regression model by collecting data from patients with lung cancer at Al-Sultan Abdullah Hospital (UiTM Hospital), Selangor. The results found that haemoptysis and chest pain has been proven to be the highest risk, among other symptoms obtained from the data analysis. It has been discovered that the H-value of 0.0 has the least measurement error, with mean square error (MSE) and root mean square error (RMSE) values of 1.455 and 1.206, respectively.

Author 1: Aliya Syaffa Zakaria

Author 2: Muhammad Ammar Shafi

Author 3: Mohd Arif Mohd Zim

Author 4: Siti Noor Asyikin Mohd Razali

Keywords: Lung cancer; high-risk symptom; fuzzy linear regression; H-value; mean square error

Paper 63: Two Phase Detection Process to Mitigate LRDDoS Attack in Cloud Computing Environment

Abstract: Distributed Denial of Service (DDoS) is a major attack carried out by attackers leveraging critical cloud computing technologies. DDoS attacks are carried out by flooding the victim servers with a massive volume of malicious traffic over a short period, Because of the enormous amount of malicious traffic, such assaults are easily detected. As a result, DDoS operations are increasingly appealing to attackers due to their stealth and low traffic rates, DDoS assaults with low traffic rates are also difficult to detect. In recent years, there has been a lot of focus on defense against low-rate DDoS attacks. This paper presents a two-phase detection technique for mitigating and reducing LRDDoS threats in a cloud environment. The proposed model includes two phases: one for calculating predicted packet size and entropy, and another for calculating the covariance vector. In this model, each cloud user accesses the cloud using the virtual machine, which has a unique session ID. This model identifies all LRDDoS assaults that take place by using different protocols (TCP, UDP, ICMP). The experiment's findings demonstrate, how the suggested data packet size, IP address, and flow behavior is used to identify attacks and prevent hostile users from using cloud services. The VM instances used by different users are controlled by this dynamic mitigation mechanism, which also upholds the cloud service quality. The results of the experiments reveal that the suggested method identifies LRDDoS attacks with excellent accuracy and scalability.

Author 1: Amrutha Muralidharan Nair

Author 2: R Santhosh

Keywords: LRDDoS attack; distance deviation; covariance vector; threshold

Paper 64: Pig Health Abnormality Detection Based on Behavior Patterns in Activity Periods using Deep Learning

Abstract: Abnormal detection of pig behaviors in pig farms is important for monitoring pig health and welfare. Pigs with health problems often have behavioral abnormalities. Observing pig behaviors can help detect pig health problems and take early treatment to prevent disease from spreading. This paper proposes a method using deep learning for automatically monitoring and detecting abnormalities in pig behaviors from cameras in pig farms based on pig behavior patterns comparison in activity periods. The approach consists of a pipeline of methods, including individual pig detection and localization, pig tracking, and behavioral abnormality analysis. From pig behaviors measured during the detection and tracking process, the behavior patterns of healthy pigs in different activity periods of the day, such as resting, eating, and playing periods, were built. Behavioral abnormalities can be detected if pigs behave differently from the normal patterns in the same activity period. The experiments showed that pig behavior patterns built in 30-minute time duration can help detect behavioral abnormalities with over 90% accuracy when applying the activity period-based approach.

Author 1: Duc Duong Tran

Author 2: Nam Duong Thanh

Keywords: Deep learning; pig tracking; behavior patterns; pig health monitoring

Paper 65: Forecasting Model of Corn Commodity Productivity in Indonesia: Production and Operations Management, Quantitative Method (POM-QM) Software

Abstract: Food is an essential ingredient needed by humans. In addition to being consumed, it can also be a valuable commodity for economic purposes through productivity of food crops. Therefore, this study aims to model the forecasting of maize productivity in Indonesia using Production and Operations Management-Quantitative Method (POM-QM) software. The data collected on productivity of corn commodities in Indonesia between 1980-2019 shows fluctuations, with both deficit and surplus periods. This study uses a time series data-based forecasting model consisting of three methods, namely Double Moving Average (DMA), Weighted Moving Average (WMA), and Single Exponential Smoothing (SES). The selection of the best model was conducted based on the Mean Absolute Deviation (MAD), Mean Square Error (MSE), and Mean Absolute Percent Error (MAPE). SES emerged as the most preferred, with a lower MAPE value of 4.913%. The predicted productivity of corn in Indonesia is estimated at 5.28 tons/ha/year, sufficient to meet consumers' demand. Therefore, governments are recommended to use this information in predicting corn productivity to meet the national demand in the future.

Author 1: Asriani

Author 2: Usman Rianse

Author 3: Surni

Author 4: Yani Taufik

Author 5: Dhian Herdhiansyah

Keywords: Forecasting model; productivity; corn commodity; POM-QM

Paper 66: Recommendation System Based on Double Ensemble Models using KNN-MF

Abstract: In today's digital environment, recommendation systems are essential as they provide personalised content to users, increasing user engagement and enhancing user satisfaction. This paper proposes a double ensemble recommendation model that combines two collaborative filtering algorithms, K Nearest Neighbour (KNN) and Matrix Factorization (MF). KNN is a neighbourhood-based algorithm that uses the similarity between users or items to make recommendations. At the same time, MF is a model-based algorithm that decomposes the user-item rating matrix into lower-dimensional matrices representing the latent user and item factors. The proposed double ensemble model uses KNN and MF to predict missing ratings matrix and combines their predictions using stacking. To evaluate the performance of the proposed ensemble model, we conducted experiments on three datasets i.e. Movielense, BookCrossing dataset and Hindi Movie dataset and compared the results with those of single algorithm approaches. The experimental results demonstrate that the double ensemble model outperforms the single algorithm approaches regarding accuracy metrics such as Mean Square Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE). The results indicate that stacked KNN and MF lead to a more robust and more accurate recommendation system.

Author 1: Krishan Kant Yadav

Author 2: Hemant Kumar Soni

Author 3: Nikhlesh Pathik

Keywords: Recommendation system; k nearest neighbour; matrix factorization; predictions using stacking; ensemble model

Paper 67: Integrating Regression Models and Climatological Data for Improved Precipitation Forecast in Southern India

Abstract: Modern technologies like Artificial Intelligence (AI) and Machine Learning (ML) replicate intelligent human behavior and offer solutions in all domains, especially for human protection and disaster management. Nowadays, in both rural and urban areas, flood control is a serious issue to overcome the vast disaster to life and property. The work proposes to identify an appropriate ML based precipitation forecast model for the flood-prone southern states of India namely Tamil Nadu, Karnataka, and Kerala which receive most precipitation using the climatological information obtained from the NASA POWER platform. The work investigates the effectiveness of ML forecasting models: Multiple Linear Regression (MLR), Support Vector Regression (SVR), Decision Tree Regression (DTR), Random Forest Regression (RFR) and Ensemble (E) learning approaches of E-MLR-SVR, E-MLR-DTR, E-MLR-RFR, E-SVR-DTR, E-SVR-RFR, E-DTR-RFR, E-MLR-SVR-DTR, E-MLR-SVR-RFR, E-MLR-DTR-RFR and E-SVR-DTR-RFR in forecasting precipitation. The E-MLR-RFR model produces improved and most precise precipitation forecast in terms of Mean Absolute Error (MAE), Mean Square Error (MSE), Root Mean Square Error (RMSE) and R2 values. A higher precipitation forecast can be used to provide early warning about the possible flood in any region.

Author 1: J. Subha

Author 2: S. Saudia

Keywords: Ensemble models; machine learning models; rainfall precipitation forecast; R-squared value

Paper 68: Recurrent Ascendancy Feature Subset Training Model using Deep CNN Model for ECG based Arrhythmia Classification

Abstract: The World Health Organization (WHO) has released a report warning of the worldwide epidemic of heart disease, which is reaching worrisome proportions among adults aged 40 and high. Heart problems can be detected and diagnosed by a variety of methods and procedures. Scientists are striving to find multiple approaches that meet the required accuracy standards. Finding the heart issue in the waveform is what an Electrocardiogram (ECG) is all about. Feature-based deep learning algorithms have been essential in the medical sciences for decades, centralising data in the cloud and making it available to researchers around the world. To promptly detect irregularities in the cardiac rhythm, manual analysis of the ECG signal is insufficient. ECGs play a crucial role in the evaluation of cardiac arrhythmias in the context of daily clinical practice. In this research, a deep learning-based Convolution Neural Network (CNN) framework is adapted from its original classification task to automatically diagnose arrhythmias in ECGs. A deep convolution network that has been used for training with most relevant feature subset is used for accurate classification. The primary goal of this research is to classify arrhythmia using a deep learning method that is straightforward, accurate, and easily deployable. This research proposes a Recurrent Ascendancy Feature Subset Training model using Deep CNN model for arrhythmia Classification (RAFST-DCNN-AC). The suggested framework is tested on ECG waveform circumstances taken from the MIT-BIH arrhythmia database. The proposed model when contrasted with the existing models exhibit better classification rate.

Author 1: Shaik Janbhasha

Author 2: S Nagakishore Bhavanam

Keywords: Feature selection; arrhythmia classification; convolution neural network; deep learning; electrocardiograms

Paper 69: Machine Learning Techniques in Keratoconus Classification: A Systematic Review

Abstract: Machine learning (ML) algorithms are being integrated into several disciplines. Ophthalmology is one field of health sector that has benefited from the advantages and capacities of ML in processing of different types of data. In a large number of studies, the detection and classification of various diseases, such as keratoconus, was carried out by analyzing corneal characteristics, in different data types (images, measurements, etc.), using ML tools. The main objective of this study was to conduct a rigorous systematic review of the use of ML techniques in the detection and classification of keratoconus. Papers considered in this study were selected carefully from Scopus and Web of Science digital databases, according to their content and to the adoption of ML methods in the classification of keratoconus. The selected studies were reviewed to identify different ML techniques implemented and the data types handled in the diagnosis of keratoconus. A total of 38 articles, published between 2005 and 2022, were retained for review and discussion of their content.

Author 1: AATILA Mustapha

Author 2: LACHGAR Mohamed

Author 3: HRIMECH Hamid

Author 4: KARTIT Ali

Keywords: Ophthalmology; corneal disease; keratoconus classification; machine learning

Paper 70: Impact and Analysis of Disease Spread in Paddy Crops using Environmental Factors with the Support of X-Step Algorithm

Abstract: India is an agriculture-based country, with paddy being the main crop cultivated on nearly half of its agricultural lands. Paddy cultivation faces numerous challenges, particularly diseases that affect crop growth and yield. Adult paddy crops are especially vulnerable to diseases caused by various factors, such as the green rice leafhopper, rice leaf folder, and brown plant leafhopper. These insects inflict damage on the paddy crops, restricting their growth and leading to significant losses. This research paper investigates the impact of environmental factors on disease spread in paddy crops, using the X-Step Algorithm for analysis. The study aims to better understand the role of environmental conditions, including air, water, and soil quality, in the development and progression of diseases in rice crops. This knowledge will help to optimize disease prevention and management strategies for improved crop yields and food security. The X-Step Algorithm, a novel machine learning algorithm, was employed to model and predict disease spread, taking into account various environmental factors. The proposed algorithm analyses images of paddy crops either manually captured or taken by sensors to evaluate disease spread and growth in paddy crops. This data-driven approach allows for more accurate and timely predictions, enabling farmers and agricultural experts to implement appropriate interventions.

Author 1: P. Veera Prakash

Author 2: Muktevi Srivenkatesh

Keywords: Paddy crops; cash crop disease; green rice leafhopper; rice leaf folder; brown plant leafhopper; x-step algorithm

Paper 71: Study of Student Personality Trait on Spear-Phishing Susceptibility Behavior

Abstract: Spear-phishing emails are an effective cyber-attack method due to the fact that the emails sent are highly personalized to look like a regular legitimate email. Recently, it was discovered that personality traits of the victim have an impact on a person's susceptibility to spear-phishing. This study aims to identify which personality traits affect spear-phishing susceptibility besides other traits such as Information Technology background, gender, and age. In addition, measure of the effectiveness of embedded training systems and see whether message framing can further help increase its effectiveness. A personality trait survey was sent to 100 participants, followed by a real-life spear-phishing simulation to measure a certain personality trait’s influence on phishing susceptibility. After a two-week period, the second round of spear-phishing emails was sent again to measure message framing effectiveness. The personality traits analysis results show that users with higher levels of Internet anxiety are less susceptible to spear-phishing emails. While the message framing did not show any significant results, the embedded training program reduced the click rate. Findings revealed that certain people are more susceptible to spear-phishing emails than others. Thus, this work can guide an institution or organizations to identify which group of people are more vulnerable to spear-phishing.

Author 1: Mohamad Alhaddad

Author 2: Masnizah Mohd

Author 3: Faizan Qamar

Author 4: Mohsin Imam

Keywords: Spear-phishing; cyber-attack; personality; trait; embedded training; message framing

Paper 72: Investigating Internet of Things Impact on e-Learning System: An Overview

Abstract: e-Learning systems have reached their peak with the revolution of smart technologies. In the past few years, the Internet of Things (IoT) has become one of the most advanced and popular technologies, affecting many different areas. Using IoT in an e-learning system is a fantastic technology that improves the e-learning system and makes it more inventive and cutting-edge. The key challenge addressed in this study is the acceptance of IoT usage in e-learning systems as well as how to improve it so that it can be utilized properly. This research concentrates on how IoT can benefit e-learning systems and how it might benefit users of e-learning systems. A comprehensive literature review was conducted to get acquainted with the important research related to IoT technology and e-learning systems through online research databases and reliable scientific journals. The first research finding is that e-learning systems need such modern techniques as IoT to enable interconnection, increase reliability, and enhance the enjoyment of the educational process. The second result is that research related to the development of new technologies like the IoT has a significant impact on enhancing the performance of new systems and bringing about positive change. This study highlights the value of IoT, particularly in e-learning systems. It aids in the development of new strategies that will improve the efficacy of e-learning systems and stimulate researchers to develop advanced technology.

Author 1: Duha Awad H. Elneel

Author 2: Hasan Kahtan

Author 3: Abdul Sahli Fakharudin

Author 4: Mansoor Abdulhak

Author 5: Ahmad Salah Al-Ahmad

Author 6: Yehia Ibrahim Alzoubi

Keywords: e-Learning system; Internet of Things (IoT); software system; education; learning process

Paper 73: Automatic Classification of Scanned Electronic University Documents using Deep Neural Networks with Conv2D Layers

Abstract: This paper proposes a novel approach for scanned document categorization using a deep neural network architecture. The proposed approach leverages the strengths of both convolutional neural networks (CNNs) and recurrent neural networks (RNNs) to extract features from the scanned documents and model the dependencies between words in the documents. The pre-processed documents are first fed into a CNN, which learns and extracts features from the images. The extracted features are then passed through an RNN, which models the sequential nature of the text. The RNN produces a probability distribution over the predefined categories, and the document is classified into the category with the highest probability. The proposed approach is evaluated on a dataset of scanned documents, where each document is categorized into one of four predefined categories. The experimental results demonstrate that the proposed approach achieves high accuracy and outperforms existing methods. The proposed approach achieves an overall accuracy of 97.3%, which is significantly higher than the existing methods' accuracy. Additionally, the proposed approach's performance was robust to variations in the quality of the scanned documents and the OCR accuracy. The contributions of this paper are twofold. Firstly, it proposes a novel approach for scanned document categorization using deep neural networks that leverages the strengths of CNNs and RNNs. Secondly; it demonstrates the effectiveness of the proposed approach on a dataset of scanned documents, highlighting its potential applications in various domains, such as information retrieval, data mining, and document management. The proposed approach can help organizations manage and analyze large volumes of data efficiently.

Author 1: Aigerim Baimakhanova

Author 2: Ainur Zhumadillayeva

Author 3: Sailaugul Avdarsol

Author 4: Yermakhan Zhabayev

Author 5: Makhabbat Revshenova

Author 6: Zhenis Aimeshov

Author 7: Yerkebulan Uxikbayev

Keywords: Deep learning; CNN; RNN; classification; image analysis

Paper 74: Combinatorial Optimization Design of Search Tree Model Based on Hash Storage

Abstract: The game search tree model usually does not consider the state information of similar nodes, which results in searching a huge state space, and there are problems such as the size of the game tree and the long solution time. In view of this, the article proposes a scheme using the idea of combinatorial optimization algorithm, which has an important application in solving the decision problem in the tree graph model. First, the special graph-theoretic structure of the point-grid game is analyzed, and the storage and search of states are optimized by designing hash functions; then, the branch delimitation algorithm is used to search the state space, and the evaluation value of repeated nodes is calculated by dynamic programming; finally, the state space is greatly reduced by combining the two-way detection search strategy. The results show that the algorithm improves decision-making efficiency and has achieved 37% and 42% final winning rate, respectively. The design provides new ideas for computational complexity problems in the field of game search and also proposes new solutions for the field of combinatorial optimization.

Author 1: Yun Liu

Author 2: Jiajun Li

Author 3: Jingjing Chen

Keywords: Combination optimization; game search algorithm; state space; transposition table

Paper 75: Skin Cancer Image Detection and Classification by CNN based Ensemble Learning

Abstract: Melanoma is accounted as a rare skin cancer responsible for a huge mortality rate. However, various imaging tests can be used to detect the metastatic spread of disease with a primary diagnosis or on clinical suspicion. Focus on melanoma detection, irrespective of its unusual occurrence, is that it is often misdiagnosed for other skin malignancies leading to medical negligence. Sometimes melanoma is detected only when the metastasis has entered the bloodstream or lymph nodes. So, effective computational strategies for early detection of melanoma are essential. There are four principal types of skin melanoma with two sub types: Superficial spreading, nodular, lentigo, lentigo maligna, Acral lentiginous, and Subungual melanoma. Amelanotic melanoma, one particular type of melanoma, exists in all kinds of skin tones. Classifications of melanoma with its classes are focused on in this research. Misclassification errors, overfitting issues and improve accuracy, the ensemble classifier models, namely Adaboost, random forest, voted ensemble, voted CNN, Boosted SVM, Boosted GMM, have been used in melanoma classification. The results of the ensemble classifier achieve high classification accuracy. However, imbalanced classification is found in all six classes of melanoma. Transfer learning and ensembled transfer learning approaches are implemented to reduce the imbalanced classification issues, and performances are analyzed. Four ML/DL models, six ensembled models, four transfer learning models, and five ensembled transfer learning models are used in this investigation. Implementation of all the 19 classifiers is analyzed using standard performance metrics such as Accuracy, Precision, recall, Mathew’s correlation coefficient, Jaccard Index, F1 measure, and Cohen’s Kappa.

Author 1: Sarah Ali Alshawi

Author 2: Ghazwan Fouad Kadhim AI Musawi

Keywords: Medical images; skin cancer; machine learning; deep learning; ensemble learning; accuracy

Paper 76: Krill Herd Algorithm for Live Virtual Machines Migration in Cloud Environments

Abstract: Green cloud computing is a modern approach that provides pay-per-use information and communication technologies with a minimal carbon footprint. Cloud computing enables users to access computing resources without the need for local servers or personal devices to operate applications. It allows businesses and developers to access infrastructure and hardware resources conveniently. Consequently, this results in a growing demand for data centers. It becomes crucial in maintaining economic and environmental sustainability as data centers use disproportionate energy. This points to sustainability and energy consumption as being important topics for research in cloud computing. This paper introduces a two-tiered VM placement algorithm. A queuing model is proposed at the first level to handle many VM requests. Models such as cloud simulation are easily implemented and validated using this model. It also provides an alternate method for allocating tasks to servers. Next, a multi-objective VM placement algorithm is proposed based on the Krill Herd (KH) algorithm. Basically, it maintains a balance between energy consumption and resource utilization.

Author 1: Hui Cao

Author 2: Zhuo Hou

Keywords: Cloud computing; migration; virtualization; energy consumption

Paper 77: Using the Term Frequency-Inverse Document Frequency for the Problem of Identifying Shrimp Diseases with State Description Text

Abstract: With the increasing demand for research on shrimp disease recognition to assist far-off farmers who need the proper assistance for their shrimp farming, shrimp disease prediction research is still in the initial stage. Most current methods utilize vision-based models, which mainly face challenges: symptom detection and image quality. Meanwhile, there are few researches which are language-based to get over the issues. In this study, we will experiment with natural language processing based on recognizing shrimp diseases; based on descriptions of shrimp status. This study provides an efficient solution for classifying multiple diseases in shrimp. We will compare different machine learning models and deep learning models (SVM, Logistic Regression, Multinomial Naive Bayes, (a4) Bernoulli Naive Bayes, Random forest, DNN, LSTM, GRU, BRNN, RCNN) in terms of accuracy and performance. The study also evaluates the TF-IDF technique in feature extraction. Data were collected for 12 types of shrimp diseases with 1,037 descriptions. Firstly, the data is preprocessed with standardised Vietnamese accent typing, tokenized words, converted to lowercase, removed unnecessary characters and stopwords. Then, TF-IDF is utilized to express the text feature weight. Machine learning-based and deep learning-based models are trained. The experimental results show that Random forest (F1-Score micro: 98%) and DNN (Validation accuracy: 84%) are the most efficient models.

Author 1: Luyl-Da Quach

Author 2: Anh Nguyen Quynh

Author 3: Khang Nguyen Quoc

Author 4: An Nguyen Thi Thu

Keywords: TF-IDF; machine learning; deep learning; CNN; shrimp disease classification

Paper 78: MoveNET Enabled Neural Network for Fast Detection of Physical Bullying in Educational Institutions

Abstract: In this article, we provide a MoveNET-based technique that we think may be used to detect violent actions. This strategy does not need high-computational technology, and it is able to put into action in a very short amount of time. Our method is comprised of two stages: first, the capture of features from photo sequences in order to evaluate body position; next, the application of an artificial neural network to activities classification in order to determine whether or not the picture frames include violent or hostile circumstances. A video aggression database consisting of 400 minutes of one individual's actions and 20 hours of videodata encompassing physical abuse, as well as 13 categories for distinguishing between the behaviors of the attacker and the victim, was created. In the end, the suggested approach was refined and validated by employing the collected dataset during the process. According to the findings, an accuracy rate of 98% was attained while attempting to detect aggressive behavior in video sequences. In addition, the findings indicate that the suggested technique is able to identify aggressive behavior and violent acts in a very short amount of time and is suitable for use in apps that take place in the real world.

Author 1: Zhadra Kozhamkulova

Author 2: Bibinur Kirgizbayeva

Author 3: Gulbakyt Sembina

Author 4: Ulmeken Smailova

Author 5: Madina Suleimenova

Author 6: Arailym Keneskanova

Author 7: Zhumakul Baizakova

Keywords: MoveNET; neural networks; skeleton; bullying; machine learning

Paper 79: Design of a Reliable Transmission Mechanism for Vehicle Data in Mobile Internet of Vehicles Driven by Edge Computing

Abstract: In order to meet the business requirements of different applications in heterogeneous, random, and time-varying mobile network environments, the design of a reliable transmission mechanism is the core problem of the mobile Internet of vehicles. The current research is mainly based on the computing power support of roadside units, and large delays and high costs are significant defects that are difficult to overcome. In order to overcome this deficiency, this paper integrates edge computing to design task unloading and routing protocol for the reliable transmission mechanism of mobile Internet of vehicles. Firstly, combined with edge computing technology, a mobile-aware edge task unloading mechanism in a vehicle environment is designed to improve resource utilization efficiency and strengthen network edge computing capacity so as to provide computing support for upper service applications; Secondly, with the support of computing power of edge task unloading mechanism, connectivity aware and delay oriented edge node routing protocol in-vehicle environment is constructed to realize reliable communication between vehicles. The main characteristics of this research are as follows: firstly, edge computing technology is introduced to provide distributed computing power, and reliable transmission routing is designed based on vehicle-to-vehicle network topology, which has prominent cost advantages and application value. Secondly, the reliability of transmission is improved through a variety of innovative technical designs, including taking the two hop range nodes as the service set search to reduce the amount of system calculation, fully considering the link connectivity state, and comprehensively using real-time and historical link data to establish the backbone link. This paper constructs measurement indicators based on delay and mobility as key elements of the computing offloading mechanism. The offloading decision is made through weighted calculation of delay estimation and computing cost, and a reasonable computing model is designed. The experimental simulation shows that the average task execution time under this model is 65.4% shorter than that of local computing, 18.4% shorter than that of cloud computing, and the routing coverage is about 6% higher than that of local computing when there are less than 60 nodes. These research and experimental results fully demonstrate that the mobile Internet of vehicles based on edge computing has good reliable transmission characteristics.

Author 1: Wenjing Liu

Keywords: Mobile network; internet of vehicles; reliable transmission; edge computing

Paper 80: Design of Intrusion Detection System using Ensemble Learning Technique in Cloud Computing Environment

Abstract: The key advantage of the cloud is that it fluidly propagates to fulfil changeable requirements and provides an environment that is repeatable and can be scaled down instantly when needed. Therefore, it is necessary to protect this cloud environment from malicious attacks such as spamming, keylogging, Denial of Service (DoS), and Distributed Denial of Service (DDoS). Among these kinds of attacks, DDoS has the capability to establish a high flood of malicious attacks on the cloud environment or Software Defined Networking (SDN) based cloud environment. Hence in this work, an ensemble based deep learning technique is proposed to detect attacks in cloud and SDN based cloud environments. Here, the ensemble model is formed by combining K-means with deep learning classifiers such as Long Short term Memory (LSTM) network, Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Gated Recurrent Unit (GRU) and Deep Neural Network (DNN). Initially, preprocessing with data cleaning and standardization is applied to the input data. Meanwhile, a random forest is implemented for extracting the minimum significant features. After that, the proposed ensemble based approach is utilized for detecting the intrusion. This approach is used to enhance the performance of the deep learning classifiers without much computational complexity. This model is trained and evaluated using two datasets as CICIDS 2018 and SDN based DDOS attack datasets. The proposed approach provides better intrusion detection performance in terms of F1 measure, precision, accuracy, and recall. By using the proposed approach, the accuracy and precision value attained is 99.685 and 0.992, respectively.

Author 1: Rajesh Bingu

Author 2: S. Jothilakshmi

Keywords: Cloud; distributed denial of service; intrusion detection; ensemble; recurrent neural network; convolutional neural network; random forest; gated recurrent unit; K-means clustering; long short term memory

Paper 81: A Novel Method for Anomaly Detection in the Internet of Things using Whale Optimization Algorithm

Abstract: The Internet of Things (IoT) is integral to human life due to its pervasive applications in home appliances, surveillance, and environment monitoring. Resource-constrained IoT devices are easily accessible to attackers due to their direct connection to the unsafe Internet. Public access to the Internet makes IoT objects more susceptible to intrusion. As the name implies, anomaly detection systems are designed to identify anomalous traffic patterns that conventional firewalls fail to detect. Effective Intrusion Detection Systems (IDSs) design faces three major problems, including handling high dimensionality, selecting a learning algorithm, and comparing entered observations and traffic patterns using a distance or similarity measure. Considering the dynamic nature of the entities involved and the limited computing resources available, more than traditional anomaly detection approaches is required. This paper proposes a novel method based on Whale Optimization Algorithm (WOA) to detect anomalies in IoT-based networks that conventional firewall systems cannot detect. Experiments are conducted on the KDD dataset. The accuracy of the proposed method is compared for classifiers such as kNN, SVM, and DT approaches. The detection accuracy rate of the proposed method is significantly higher than that of other methods for DoS, probing, normal attacks, R2L attacks, and U2R attacks compared to other methods. This method shows an impressive increase in accuracy when detecting a wide range of malicious activities, from DoS, probing, and privilege escalation attacks, to remote-to-local and user-to-root attacks.

Author 1: Zhihui Zhu

Author 2: Meifang Zhu

Keywords: Internet of things; anomaly detection; intrusion detection; firewall; whale optimization algorithm; accuracy

Paper 82: Autonomous Path Planning for Industrial Omnidirectional AGV Based on Mechatronic Engineering Intelligent Optical Sensors

Abstract: With the rapid development of modern industry, the application of automated mechanical and electronic technology is gradually increasing, and the research on automatic path planning is also receiving increasing attention. In this environment of rapid technological progress, rapid growth of the knowledge economy, and fierce competition, industrial intelligence has become an indispensable part of social development. Industrial Automated Guided Vehicle (AGV) has put forward higher requirements for the application of automatic control technology in the planning and research of autonomous path planning. Autonomous path planning with AGV as the service object is currently the most widely used direction in industrial production processes, with the best development prospects and the highest market demand. Optimizing autonomous path planning for AGV is of great significance in promoting the process of industrial modernization and improving industrial production efficiency. In order to solve the problems of low path planning efficiency, excessive reliance on the rich experience and subjective judgment of relevant personnel, and excessive consumption of path planning costs in traditional AGV omnidirectional autonomous path planning, this article attempted to introduce sensor technology to conduct in-depth research on AGV omnidirectional automatic path planning. Based on intelligent optical sensors and combined with ant colony algorithm, the autonomous path planning model for AGV was optimized, and an innovative AGV omnidirectional autonomous path planning model application experiment was conducted in two industrial production enterprises in a certain region. Comparative analysis of experimental data showed that the innovative AGV omnidirectional autonomous path planning model studied in this article had an average improvement of about 17.8% in four evaluation indicators compared to traditional AGV omnidirectional autonomous path planning models.

Author 1: Yuanyuan Pan

Keywords: Smart machinery; optical sensors; industrial development; autonomous path planning

Paper 83: A Study on the Evaluation Model of In-depth Learning for Oral English Learning in Online Education

Abstract: The trend of globalization in the world is becoming increasingly frequent, and people from different regions are communicating more closely. Therefore, the demand for a second language is constantly expanding, accelerating the development of the field of English oral evaluation and also accelerating the development of online education. The study proposes a text priori based oral evaluation model, which is based on the Transformer model and uses target phonemes as input to the Decoder. The model successfully predicts the relationship between actual pronunciation and error labels. At the same time, a self-supervised oral evaluation model with accent is constructed, which simulates the training process of misreading data by calculating semantic distance. The experimental results show that when the training set ratio reaches its maximum in the Speed Ocean dataset and the L2 Arctic dataset, the F1 values of the proposed method are 0.612 and 0.596, respectively; the length of the target phoneme has a smaller impact on this model compared to other models. Experiments have shown that the proposed deep learning method can alleviate deployment difficulties, directly optimize the effectiveness of oral evaluation, provide more accurate feedback, and also provide users with a better learning experience. This has practical significance for the development of the field of oral evaluation.

Author 1: Yanli Ge

Keywords: Spoken English; online education; transformer model; deep learning; evaluation model

Paper 84: Attribute-based Access Control Model in Healthcare Systems with Blockchain Technology

Abstract: Blockchain and the healthcare sector have a serious concern with context to scalability, which has a challenge of converting arbitrary values to fixed values. The transfer of arbitrary data coming from diverse resources has another point of concern in the blockchain. In this paper, the author proposed a model that will receive data from diverse sources and will convert it to a fixed type of value. The paper also proposes an access control scheme with various permission and consensus level protocols which will allow a reduction in block size with respect to scalability. The consensus level will allow access to the individual or a group of users and the permission level with respect to each block via considering the access granted to nodes of the blockchain. The addition of various permission and consensus levels will allow only a restricted type of data to pass the model. Once the data is verified and approved by various levels, then the data is all set to be part of the blockchain. The paper introduces a model where the time taken to create a new hash is 0.15625 microseconds. A total number of 64 transactions taken from the data set where the throughput is calculated for individual access are considered. After applying the formula, the calculated throughput is 32.5 microseconds. By the lighter block size data can be made available to the patients. The research is for the patients so they can keep track of their medical history and the deaths due to overdose of the medicines can be reduced.

Author 1: Prince Arora

Author 2: Avinash Bhagat

Author 3: Mukesh Kumar

Keywords: Blockchain; healthcare; permission level; consensus level; scalability

Paper 85: A New Design of Optical Logic Gates with Transverse Electric and Magnetic

Abstract: This paper presents a new design of optical NOR and XNOR logic gates using a two-dimensional-hexagonal photonic crystal (2D-HPhC) that allows for both Transverse Electric (TE) and Transverse Magnetic (TM) polarization modes. The structure is very small in size and has a low delay time. The design includes three inputs (A, B, C) and one output (Q) waveguide, with the NOR gate having a delay period of 0.75 ps and the XNOR gate having a delay period of 0.9 ps. The contrast ratio between the input and output for both gates is 7-8 dB. The XNOR gate has an optimum transmission signal rate of T = 96%. The purpose of the structure is to use a reference input to create the fundamental logic gates NOR and XNOR by adjusting the signal phase angle.

Author 1: Lili Liu

Author 2: Haiquan Sun

Author 3: Lishuang Hao

Author 4: Cailiang Chen

Keywords: Photonic crystal; hexagonal lattice; NOR; XNOR; transverse electric; transverse magnetic

Paper 86: Weapons Detection System Based on Edge Computing and Computer Vision

Abstract: Early detection of armed threats is crucial in reducing accidents and deaths resulting from armed conflicts and terrorist attacks. The most significant application of weapon detection systems would be found in public areas such as airports, stadiums, central squares, and on the battlefield in urban or rural conditions. Modern surveillance and control systems of closed-circuit television cameras apply deep learning and machine learning algorithms for weapons detection on the base of cloud architecture. However, cloud computing is inefficient for network bandwidth, data privacy and slow decision-making. To address these issues, edge computing can be applied, using Raspberry Pi as an edge device with the EfficientDet model for developing the weapons detection system. The image processing results are transmitted as a text report to the cloud platform for further analysis by the operator. Soldiers can equip themselves with the suggested edge node and headphones for armed threat notifications, plugged into augmented reality glasses for visual data output. As a result, the application of edge computing makes it possible to ensure data safety, increase the network bandwidth and provide the device operation without the internet. Thus, an independent weapon detection system was developed that identifies weapons in 1.48 seconds without the Internet.

Author 1: Zufar R. Burnayev

Author 2: Daulet O. Toibazarov

Author 3: Sabyrzhan K. Atanov

Author 4: Hüseyin Canbolat

Author 5: Zhexen Y. Seitbattalov

Author 6: Dauren D. Kassenov

Keywords: Internet of Things; gun recognition; edge device; Raspberry pi; military systems control; network analytic

Paper 87: Distributed Cooperative Control for Multi-UAV Flying Formation

Abstract: The problem of collaborative pattern tracking in multi-agent systems (MAS) like unmanned aerial vehicles (UAV) is investigated in this article. First, a new method for distributed consensus is constructed inside the framework of the leader-following approach for second-order nonlinear MAS. The technique canceled the chattering effect observed in the conventional sliding mode-based control protocols by transmitting smooth input signals to agents' control channels. Second, a novel formation framework is proposed to accomplish three-dimensional formation tracking by including consensus procedures in the formation dynamics model. This will allow for formation tracking in all three dimensions. The Lyapunov theory provides evidence demonstrating the proposed protocols' stability and convergence. Numerical simulations have been carried out to prove the proposed algorithms' effectiveness.

Author 1: Belkacem Kada

Author 2: Abdullah Y.Tameem

Author 3: Ahmed A. Alzubairi

Author 4: Uzair Ansari

Keywords: Formation control; distributed consensus; multi-agent systems; multiple-UAV

Paper 88: An Artificial Intelligent Methodology-based Bayesian Belief Networks Constructing for Big Data Economic Indicators Prediction

Abstract: Economic indicator prediction in big data requires treating all random variables as an independent set of selective values and used as a discriminative method for classification tasks. A Bayesian network is a popular graphical representation approach for modeling probabilistic dependencies and causality among a set of random variables to incorporate a huge amount of human expert knowledge about the problem of interest involving diagnostic reasoning of big data. In our study, we set out to construct the Bayesian networks using the standard error for a least-squares linear regression (STE) and the domain knowledge from the literature in the field for predicting the big data economy prediction. The experimental results show that the proposed STE baseline provided us with an accuracy of 20% to 58% in seven out of eight regions, including the aggregate for “World”. In comparison, the Bayesian Networks generated by our first Domain Knowledge Model improved accuracy from 54% to 75% in the same regions.

Author 1: Adil Al-Azzawi

Author 2: Fernando Torre Mora

Author 3: Chanmann Lim

Author 4: Yi Shang

Keywords: STE; Bayesian networks; domain knowledge; discriminative methods; economic forecasting

Paper 89: The Application of Virtual Technology Based on Posture Recognition in Art Design Teaching

Abstract: With the development of virtual technology, posture recognition technology has been integrated into virtual technology. This new technology allows users to further understand and observe the activities carried out in life scenes based on their original observation of the external world. And it enables them to make intelligent decisions. Existing posture recognition cannot meet the requirements of precise positioning in virtual environments. Therefore, a two-stage three-dimensional pose recognition model is proposed. The experiment illustrates that the three-dimensional gesture recognition performance is excellent. In addition, under the ablation experiment, the error accuracy of the research model decreased by more than 5 mm, and the overall error accuracy decreased by 10%. In the P-R curve, the accuracy rate of the model reaches 0.741, and the recall rate reaches 0.65. When conducting empirical analysis, the virtual posture disassembly action is complete; the disassembly error is less than 5%, and the disassembly error accuracy is good. The fit degree of the leg bending amplitude reaches over 96%, and the fit degree of the arm bending amplitude reaches over 95%. When the model is applied to actual teaching, the overall satisfaction score of teachers and students reaches 94.6 points. This has effectively improved the teaching effect of art design and is of great significance to the development of education in China.

Author 1: Qinyan Gao

Keywords: Posture recognition; deep learning; art design; time convolutional network; VR

Paper 90: Business Data Analysis Based on Kissmetric in the Context of Big Data

Abstract: The kissmetric data analysis model can be used for the analysis and research of business data, and the focused research method in this model is cluster analysis. To realize the effective application of Kissmetric data analysis model, the focused method is improved in the experiment. An improved hierarchical clustering algorithm generated by splitting stage and merging stage is proposed in the experiment, and then the algorithm is combined with density clustering method while considering noise point processing to achieve automatic determination of clustering centers and improvement of clustering effect. In different dimensions, the highest F-measure index and ARI values of the hybrid clustering method are 0.997 and 0.998, respectively. In different numbers of classes of the dataset, the highest F-measure index and ARI values of the hybrid clustering method are 1.000 and 0.999, respectively. The mean accuracy and mean-variance were 95.94% vs. 5.89%, 94.72% vs. 0.57%, 89.72% vs. 4.97%, 87.45% vs. 5.53%, 93.83% vs. 5.76%, and 88.43% vs. 5.40 %, respectively. The mean and mean squared deviation of hybrid clustering method’s accuracy was 89.71% vs. 6.17% and 88.85% vs. 0.33% when dealing with the real datasets 7 and 8, respectively. The quality and stability of the clustering results of the hybrid clustering method are better. Compared with other clustering methods, the accuracy and stability of this method are higher and have certain superiority.

Author 1: Kan Wang

Keywords: Big data; kissmetric; data analysis; density clustering; hierarchical clustering

Paper 91: From Monolith to Microservice: Measuring Architecture Maintainability

Abstract: The migration of monolithic applications to the cloud is a popular trend, with microservice architecture being a commonly targeted architectural pattern. The motivation behind this migration is often rooted in the challenges associated with maintaining legacy applications and the need to adapt to rapidly changing business requirements. To ensure that the migration to microservices is a sound decision for enhancing maintainability, designers must carefully consider the underlying factors driving this software architecture migration. This study proposes a set of software architecture metrics for evaluating the maintainability of microservice architectural designs for monolith to microservice architecture migration. These metrics consider various factors, such as coupling, complexity, cohesion, and size, which are crucial for ensuring that the software architecture remains maintainable in the long term. Drawing upon previous product quality models that share similar design properties with microservice, we have derived maintainability metrics that can help measure the quality of microservice architecture. In this work, we introduced our first version of structural metrics for measuring the maintainability quality of microservice architecture concerning its cloud-native characteristics. This work allows us to get early feedback on proposed metrics before a detailed evaluation. With these metrics, designers can measure their microservice architecture quality to fully leverage the benefits of the cloud environment, thus ensuring that the migration to microservice is a beneficial decision for enhancing the maintainability of their software architecture applications.

Author 1: Muhammad Hafiz Hasan

Author 2: Mohd. Hafeez Osman

Author 3: Novia Indriaty Admodisastro

Author 4: Muhamad Sufri Muhammad

Keywords: Monolith; cloud migration; software architecture; design quality; maintainability; quality metric

Paper 92: Improved 3D Rotation-based Geometric Data Perturbation Based on Medical Data Preservation in Big Data

Abstract: With the rise in technology, a huge volume of data is being processed using data mining, especially in the healthcare sector. Usually, medical data consist of a lot of personal data, and third parties utilize it for the data mining process. Perturbation in health care data highly aids in preventing intruders from utilizing the patient’s privacy. One of the challenges in data perturbation is managing data utility and privacy protection. Medical data mining has certain special properties compared with other data mining fields. Hence, in this work, the machine learning (ML) based perturbation approach is introduced to provide more privacy to healthcare data. Here, clustering and IGDP-3DR processes are applied to improve healthcare privacy preservation. Initially, the dataset is pre-processed using data normalization. Then, the dimensionality is reduced by SVD with PCA (singular value decomposition with Principal component analysis). Then, the clustering process is performed by IFCM (Improved Fuzzy C means). The high-dimensional data are divided into several segments by IFCM, and every partition is set as a cluster. Then, improved Geometric Data Perturbation (IGDP) is used to perturb the clustered data. IGDP is a combination of GDP with 3D rotation (3DR). Finally, the perturbed data are classified using a machine learning (ML) classifier - kernel Support Vector Machine- Horse Herd Optimization (KSVM-HHO) to classify the perturbed data and ensure better accuracy. The overall evaluation of the proposed KSVM-HHO is carried out in the Python platform. The performance of the IGDP-KSVM-HHO is compared over the two benchmark datasets, Wisconsin prognostic breast cancer (WBC) and Pima Indians Diabetes (PID) dataset. For the WBC dataset, the proposed method obtains an overall accuracy of 98.08% perturbed data, and for the PID dataset, the proposed method obtains an overall accuracy of 98.04%.

Author 1: Jayanti Dansana

Author 2: Manas Ranjan Kabat

Author 3: Prasant Kumar Pattnaik

Keywords: Data mining; privacy; health care data; machine learning; perturbation; improved fuzzy c-means; horse herd optimization; kernel based support vector machine

Paper 93: An Analysis of Bias in Facial Image Processing: A Review of Datasets

Abstract: Facial image processing is a major research area in digital signal processing. According to recent studies, most commercial facial image processing systems are prejudiced by bias towards specific races, ethnicities, cultures, ages, and genders. In some circumstances, bias may be traced back to the algorithms employed, while in others, bias can be elicited from the insufficient representations in datasets. This study tackles bias based on insufficient representations in datasets. To tackle this, the research undertakes an exploratory review in which the context of facial image dataset is analyzed to thoroughly examine the rate of bias. Facial image processing systems are developed using widely publicly available datasets since generating datasets are costly. However, these datasets are strongly biased towards Whites and Asians, and other geo-diversity such as indigenous Africans are underrepresented. In this study, 40 large publicly accessible facial image data sets were examined. The races of the datasets used for this study were visualized using the t-distributed Stochastic Neighbor Embedding (t-SNE) visualization method. Then, to measure the geo-diversity and rate of bias of the dataset, k-means clustering, principal component analysis (PCA) and the Oriented FAST and Rotated BRIEF (ORB) feature extraction techniques were used. The findings from this study indicate that these datasets seem to exhibit an obvious ethnicity representation bias, particularly for native African facial images; as a result, additional African indigenous datasets are required to reduce the bias currently present in the most publicly available facial image datasets.

Author 1: Amarachi M. Udefi

Author 2: Segun Aina

Author 3: Aderonke R. Lawal

Author 4: Adeniran I. Oluwarantie

Keywords: Digital signal processing; facial image processing; bias, geo-diversity; facial image datasets; k-means clustering; principal component analysis

Paper 94: Presenting a Planning Model for Urban Waste Transportation and Selling Recycled Products with a Green Chain Approach

Abstract: The growing amount of municipal solid waste (MSW) is a significant issue, especially in large urban areas with inadequate landfill capacities and ineffective waste management systems. Several supply chain options exist for implementing an MSW management system; however, numerous technical, economic, environmental, and social factors must be evaluated to determine the optimal solution. This research aims to illustrate the difficulty of urban solid waste management in a network of supply chains with several levels. Hence, a mathematical model is implemented as a mixed integer linear programming problem that encompasses a variety of functions, comprising trash collection in cities, waste separation in sorting facilities, waste treatment in industries, and waste transportation between processing facilities. In addition, given the significance of urban solid waste management to environmental concerns, we are attempting to model the problem using a green approach. The purpose of the model proposed in this article is to determine the optimal distribution of waste among all units and maximize the net profit of the entire supply chain, along with a green approach. A case study has been undertaken to evaluate the efficacy and efficiency of the suggested model, which is utilized to solve the numerical problem with GAMS software and the grasshopper metaheuristic algorithm. The findings indicate that integrating municipal solid waste can yield economic and environmental benefits.

Author 1: Baoqing Ju

Keywords: Planning model; urban waste transportation; recycled products; green supply chain

Paper 95: Improved Tuna Swarm-based U-EfficientNet: Skin Lesion Image Segmentation by Improved Tuna Swarm Optimization

Abstract: Skin cancers have been on an upward trend, with melanoma being the most severe type. A growing body of investigation is employing digital camera images to computer-aided examine suspected skin lesions for cancer. Due to the presence of distracting elements including lighting fluctuations and surface light reflections, interpretation of these images is typically difficult. Segmenting the area of the lesion from healthy skin is a crucial step in the diagnosis of cancer. Hence, in this research an optimized deep learning approach is introduced for the skin lesion segmentation. For this, the EfficientNet is integrated with the UNet for enhancing the segmentation accuracy. Also, the Improved Tuna Swarm Optimization (ITSO) is utilized for adjusting the modifiable parameters of the U-EfficientNet to minimize the information loss during the learning phase. The proposed ITSU-EfficientNet is assessed based on various evaluation measures like Accuracy, Mean Square Error (MSE), Precision, Recall, IoU, and Dice Coefficient and acquired the values are 0.94, 0.06, 0.94, 0.94, 0.92 and 0.94 respectively.

Author 1: Khaja Raoufuddin Ahmed

Author 2: Siti Zura A Jalil

Author 3: Sahnius Usman

Keywords: Skin lesion; skin cancer; segmentation; deep learning model; optimization; EfficientNet; Unet

Paper 96: Analysis and System Construction of ALSTM-LSTM Model-based Sports Jumping Rope Movement

Abstract: Computer technology's maturity has enabled intelligent and interactive sports training. Jumping rope test in secondary school faces difficulties due to bulky testing equipment and inefficient data measurement. An ALSTM-LSTM model based on visual human posture estimation is proposed for motion system analysis. Joint pose features are fused through LSTM, and the attention mechanism assigns weights to feature sequences to achieve motion recognition, considering the data's multidimensional and hierarchical nature. The model’s precision value is 95.83. Its average accuracy is much higher than LSTM, ML-KNN, and RSN models. Additionally, the model has 95.2% accuracy in localizing jump rope stance movements with low data consumption. The model can improve the accuracy of the analysis of the jump rope sport’s posture based on the characteristics of human movement, and inspire new technical tools for teaching instruction.

Author 1: Peng Su

Author 2: Zhipeng Li

Author 3: Weiguo Li

Author 4: Yongli Yang

Keywords: ALSTM-LSTM model; jumping rope exercise; Sports; human posture estimation algorithm; attention mechanisms

Paper 97: A Novel Label Propagation Method for Community Detection Based on Game Theory

Abstract: Community is a mesoscopic feature of the multi-scale phenomenon of complex networks, which is the bridge to revealing the formation and evolution of complex networks. Due to high computational efficiency, label propagation becomes a topic of considerable interest within community detection, but its randomness yet produces serious fluctuations. Facing the inherent flaws of label propagation, this paper proposes a series of solutions. Firstly, this paper presents a heuristic label propagation algorithm named Label Propagation Algorithm use Cliques and Weight (LPA-CW). In this algorithm, labels are expanded from seeds and propagated based on node linkage index. Seeds are produced from complete subgraph, and node linkage index is related to neighboring nodes. This method can produce competitive modularity Q but not Normalized Mutual Information (NMI), and compensate with existing methods, such as Stepping Community Detection Algorithm based on Label Propagation and Similarity (LPA-S). Secondly, in order to combine the advantages of different algorithms, this paper introduces a game theory framework, design the profit function of the participant algorithms to attain Nash equilibrium, and build an algorithm integration model for community detection (IA-GT). Thirdly, based on the above model, this presents an algorithm, named Label Propagation Algorithm based on IA-GT model (LPA-CW-S), which integrates LPA-CW and LPA-S and solves the incompatibility between modularity and NMI. Fully tested on both computer-generated and real-world networks, this method gives better results in indicators such as modularity and NMI than existing methods, effectively resolving the contradiction between the theoretical community and the real community. Moreover, this method significantly reduces the randomness and runs faster.

Author 1: Mengqin Ning

Author 2: Jun Gong

Author 3: Zhipeng Zhou

Keywords: Community detection; label propagation; node linkage; complete subgraph; game theory

Paper 98: Intelligent Brake Controller Based on Intelligent Highway Signs to Avoid Accidents on Algerian Roads

Abstract: Despite the considerable efforts of the Algerian authorities to reduce the high number of accidents, therefore, fatalities on the country’s roads, the problem persists. To address this worrying situation, it is necessary to adopt new technologies and approaches that can assist Algerian drivers forcing them to comply with driving rules, thus putting an end to this concerned issue. This research aims to primarily assist Algerian drivers in reducing the mortality rate, which is primarily caused by speeding and poor road conditions, including outdated and inadequate road signs. To achieve this objective, a complete system consisting of two complementary subsystems: intelligent traffic signs and an interactive and smart speed limiter, has been proposed. Existing projects in this field have shown deficiencies, particularly in the context of real-time critical systems. This work offers improved precision, real-time responsiveness, adaptability to changes, and reduced infrastructure dependency compared to existing solutions. The new approach has been tested with the SUMO simulator, and a prototype based on Arduino cards has been developed approving its feasibility. However, the results obtained from this study demonstrate that the proposed system can significantly reduce the mortality rate on Algerian roads.

Author 1: AHMED MALEK Nada

Author 2: BOUDOUR Rachid

Keywords: Intelligent Transportation Systems (ITS); Intelligent System Adaptation (ISA); road accidents; intelligent road signalization; decision

Paper 99: Experimentation on Iterated Local Search Hyper-heuristics for Combinatorial Optimization Problems

Abstract: Designing effective algorithms to solve cross-domain combinatorial optimization problems is an important goal for which manifold search methods have been extensively investigated. However, finding an optimal combination of perturbation operations for solving cross-domain optimization problems is hard because of the different characteristics of each problem and the discrepancies in the strengths of perturbation operations. The algorithm that works effectively for one problem domain may completely falter in the instances of other optimization problems. The objectives of this study are to describe three categories of a hyper-heuristic that combine low-level heuristics with an acceptance mechanism for solving cross-domain optimization problems, compare the three hyper-heuristic categories against the existing benchmark algorithms and experimentally determine the effects of low-level heuristic categorization on the standard optimization problems from the hyper-heuristic flexible framework. The hyper-heuristic categories are based on the methods of Thompson sampling and iterated local search to control the perturbation behavior of the iterated local search. The performances of the perturbation configurations in a hyper-heuristic were experimentally tested against the existing benchmark algorithms on standard optimization problems from the hyper-heuristic flexible framework. Study findings have suggested the most effective hyper-heuristic with improved performance when compared to the existing hyper-heuristics investigated for solving cross-domain optimization problems to be the one with a good balance between “single shaking” and “double shaking” strategies. The findings not only provide a foundation for establishing comparisons with other hyper-heuristics but also demonstrate a flexible alternative to investigate effective hyper-heuristics for solving complex combinatorial optimization problems.

Author 1: Stephen A. Adubi

Author 2: Olufunke O. Oladipupo

Author 3: Oludayo O. Olugbara

Keywords: Combinatorial optimization; heuristic algorithm; heuristic categorization; local search; Thompson sampling

Paper 100: Serious Game Design Principles for Children with Autism to Facilitate the Development of Emotion Regulation

Abstract: Autism spectrum disorder (ASD) is a deficit-driven neurodevelopmental condition in three areas, which are social interactions, communication, and the presence of restricted interests and repetitive behaviours. Children with autism mainly suffer from emotional disturbance that emerges as meltdowns, tantrums, and aggression, increasing the risk of developing mental health issues. Several studies have assessed the use of serious games in helping children with autism enhance their communication, learning, and social skills. Significantly, these serious games focus on the strengths and weaknesses of the disorder to establish a comfortable and controlled environment that is able to support children with autism. However, there is still a lack of evidence in studies exploring the use of serious games for children with autism to facilitate the development of emotion regulation. The aim of this study is to consolidate and propose a new serious game design principle for children with autism to facilitate the development of emotion regulation. The target age of the children involved in this study ranged between 6 and 12. A review of previous literature on serious game design principles was conducted. More than 70 articles related to serious games for children with autism were analysed using thematic analysis. This study found 16 elements that influenced the designing and developing process of creating a serious game for children with autism. It has been organised and categorised into five attributes (user, game objectives, game elements, game aesthetics, and player experience). Certainly, this study demonstrates the needs and requirements of children with autism when designing serious games.

Author 1: Nor Farah Naquiah Mohamad Daud

Author 2: Muhammad Haziq Lim Abdullah

Author 3: Mohd Hafiz Zakaria

Keywords: Autism spectrum disorders; serious game; emotion regulation; serious game design principles

Paper 101: A Novel Approach for an Outdoor Oyster Mushroom Cultivation using a Smart IoT-based Adaptive Neuro Fuzzy Controller

Abstract: An automatic environment control systems for greenhouses are turning to be very significant because of food demand, and rise in temperature and population of the world. This article proposes to design and implement a low cost, robust and water efficient autonomous smart internet of things (IoT) system to monitor and control the temperature, and humidity of an outdoor oyster mushroom growing unit. The IoT-based control system involves DHT22 sensors, ESP32 controller and actuators (water pump and cooling fan) to facilitate the adequate amount of air for circulation to maintain temperature and water to maintain humidity inside an outdoor oyster mushroom growing unit as per its requirement. A real working prototype is developed and implemented on integrating fuzzy inference system (FIS) in ESP32 controller using Arduino C with the help of its integrated design environment. The FIS is designed to calculate the switching on/off time of water pump and cooling fan on sensing current temperature, and humidity inside oyster mushroom unit with respect to ambient temperature, and humidity respectively. The prototype provides inside temperature, humidity, ambient temperature, ambient humidity, water pump time and fan time on Thing-speak platform in real time. Furthermore, the data is used for design and simulation of Adaptive Neuro Fuzzy Inference Controller for an outdoor oyster mushroom growing unit in MATLAB/Simulink to improve the performance of the system. The practical applicability of the proposed ANFIS controller over FIS Controller and industrial PID Controller is shown by simulation findings with use of experimental data. The system reduces water use as well as an extremely extraordinary administration required for monitoring the mushroom unit. In addition, it increases robustness of the system.

Author 1: Dakhole Dipali

Author 2: Thiruselvan Subramanian

Author 3: G Senthil Kumaran

Keywords: Precision agriculture; adaptive neuro fuzzy inference system; fuzzy inference system; oyster mushroom cultivation; internet of things

Paper 102: Hybrid Particle Swarm Optimization-based Modeling of Wireless Sensor Network Coverage Optimization

Abstract: To address the problem of insufficient coverage of WSN and poor network coverage in obstacle environments, the study proposes an improved particle swarm optimization (PSO) combined with a hybrid grey wolf algorithm. The speed and position of the PSO particle's search for superiority are enhanced through the guiding nature of the superior wolf in the grey wolf optimization (GWO), thus the convergence speed and search precision are improved. Based on this, the study applies the improved PSO to a wireless sensor networks (WSO) coverage optimization model and uses model comparison to test the effectiveness and superiority of the algorithm. According to the results, the node network coverage of PSO, genetic algorithm (GA), data envelopment analysis (DEA), GWO, and grey wolf particle swarm optimization (GWPSO) reach 85.97%, 87.24%, 88.76%, 89.31%, and 91.05% respectively in the trapezoidal obstacle environment. And the node network coverage of the research-designed GWPSO algorithm reaches the highest value of its kind. This shows that the research-designed GWPSO has superior performance in the optimization control of sensor coverage deployment compared with similar algorithms. The design provides a new path for optimizing wireless sensor node network coverage.

Author 1: Guangyue Kou

Author 2: Guoheng Wei

Keywords: Particle swarm optimization; wireless sensor networks; network coverage; grey wolf optimization; grey wolf particle swarm optimization

Paper 103: Effect of Multi-SVC Installation for Loss Control in Power System using Multi-Computational Techniques

Abstract: Flexible AC Transmission Systems (FACTs) play a vital role in minimizing the power losses and improving voltage profile in power transmission system. These increase the real power transfer capacity of the system. However, optimal location of sizing of the FACTs devices determines the extent of benefits provided by the FACTs devices to the transmission system. Non-optimal solution in terms of the location and sizing may possibly lead to under-compensation or over-compensation phenomena. Thus, a robust optimization is a priori for optimal solution achievement. This paper presents a study on the effect on multi static VAR compensators (SVC) installation for loss control in power system using evolutionary programming (EP), artificial immune system (AIS) and immune evolutionary programming (IEP). The objective is to minimize the real power loss transmission and improve the voltage profile of the transmission power system. The study reveals that installation of multi-units SVC significantly reduces the power loss and increases the voltage profile of the system, validated on the IEEE 30-Bus Reliability Test System (RTS).

Author 1: N. Balasubramaniam

Author 2: N. A. M. Kamari

Author 3: I. Musirin

Author 4: A. A. Ibrahim

Keywords: Flexible AC Transmission Systems (FACTs); Shunt VARs Compensators (SVCs); Evolutionary Programming (EP); Artificial Immune System (AIS); Immune Evolutionary Programming (IEP)

Paper 104: Economic Development Efficiency Based on Tobit Model: Guided by Sustainable Development

Abstract: At present, in resource-based regions in China, it has been seriously restricted in the harmonious growth of green economy (GE) and environment. To find and solve the problems that affect the quality of regional GE development, the study took Xinjiang, a resource-based province, as the research object. With the data of 14 prefectures and cities in Xinjiang from 2017 to 2022, an evaluation model for the efficiency of GE development based on DEA-Tobit was constructed. Data envelopment analysis (DEA) measures the spatial autocorrelation and distribution characteristics of GE development efficiency in various prefectures and cities in Xinjiang. The influencing factors were analyzed by using Tobit model. From the empirical results, there are obvious differences in the spatial distribution of GE development among various prefectures and cities in Xinjiang. The average value is 0.7289, the highest value is 1, and the lowest value is 0.3684, with a difference of 0.6316. The efficiency values of GE in the seven regions R1, R2, R4, R6, R7, R9, and R13 have reached 1, and DEA is effective. Based on the global and local Moran index, it can be seen that there is no obvious spatial correlation between the development efficiency of green economy in the cities and cities of Xinjiang, and the absolute value of its coefficient is not more than 0.5. From the results of the Tobit model, there are still areas for raising the efficiency of GE development in most regions of Xinjiang. Based on the established DEA-Tobit GE development efficiency evaluation model, this study proposes targeted development strategies for improving the efficiency of GE development in Xinjiang.

Author 1: Ming Liu

Keywords: Tobit model; DEA; economic development efficiency; Moran index

Paper 105: A Proposed Approach for Motif Finding Problem Solved on Heterogeneous Cluster with Best Scheduling Algorithm

Abstract: The Motif Finding Problem (MFP) is the problem of finding patterns in sequences of DNA. This paper discusses and presents an enhanced scheduling approach to solve the motif problem on the Heterogeneous Cluster by making a comparison between exact algorithms. The method that was followed is to analyze several exact algorithms, compare them within specific points to measure, and improve performance by comparing the number of devices and peripheral units used in every situation and running time in every method. Our experimental results show that the use of the scheduling approach that use different algorithms on Heterogeneous Cluster make a significant difference in the speed of completing the problem and in a shorter record time with less resources, and that this proposed approach is more effective than the traditional method of distributing tasks to solve the motif problem.

Author 1: Abdullah Barghash

Author 2: Ahmed Harbaoui

Keywords: Motif finding problem; scheduling algorithm; het-erogeneous; high-performance computing

Paper 106: Fruit Classification using Colorized Depth Images

Abstract: Fruit classification is a computer vision task that aims to classify fruit classes correctly, given an image. Nearly all fruit classification studies have used RGB color images as inputs, a few have used costly hyperspectral images, and a few classical ML-based have used colorized depth images. Depth images have apparent benefits such as invariance to lighting, less storage requirement, better foreground-background separation, and more pronounced curvature details and object edge discontinuities. However, the use of depth images in CNN-based fruit classification remains unexplored. The purpose of this study is to investigate the use of colorized depth images in fruit classification with four CNN models, namely, AlexNet, GoogleNet, ResNet101, and VGG16, and compare their performance and computational efficiency, as well as the impact of transfer learning. Depth images of apple, orange, mango, banana and rambutan (Nephelium Lappaceum) were manually collected using a depth sensor with sub-millimeter accuracy and subjected to jet, uniform, and inverse colorization to produce three sets of dataset. Results show that depth images can be used to train CNN models for fruit classification with ResNet101 achieving the best accuracy of 96%on the inverse dataset. It achieved 100% accuracy after transfer learning. GoogleNet showed the most significant improvement after transfer learning on the uniform dataset, at 12.27%. It also exhibited the lowest training and inference times. The results show the potential use of depth images for fruit classification and similar computer vision tasks.

Author 1: Dhong Fhel K. Gom-os

Keywords: Fruit classification; depth image; depth colorization; CNN; transfer learning

Paper 107: From Phishing Behavior Analysis and Feature Selection to Enhance Prediction Rate in Phishing Detection

Abstract: Phishing incidents have captured the attention of security experts and end users in recent years as they have become more frequent, widespread, and sophisticated. The researchers offered a variety of strategies for detecting phishing attacks. Over time, these approaches suffer from insufficient performance and the inability to identify zero attacks. One of the limitations with these methods is that phishing techniques are constantly evolving, and the proposed methods are not keeping up, making it a hard nut to crack. The objective of this research is to develop a URL phishing detection model that can demonstrate its robustness against constantly changing attacks. One of the most significant contributions of this paper is the selection of a novel combination of features based on literal and recent phishing behavior analysis. This makes the model competent sufficient to recognize zero attacks and able to adjust to changes in phishing attacks. Furthermore, eleven machine learning classification techniques are utilized for classification tasks and comparative objectives. Moreover, three datasets with different instance distributions were constructed at different times for the model’s initial construction and evaluation. Several experiments were carried out to investigate and evaluate the proposed model’s performance, effectiveness, and robustness. The experiments’ findings demonstrated that the GaussianNB method is the most durable, capable of maintaining performance even in the absence of retraining. Additionally, the LightGBM, Random Forest, and GradientBoost algorithms had the highest levels of performance, which they were able to maintain by routinely retraining the model with newer types of attacks. Models that employed these three suggested algorithms outperformed other current detection models with an average accuracy of about 99.7%, making them promising.

Author 1: Asmaa Reda Omar

Author 2: Shereen Taie

Author 3: Masoud E.Shaheen

Keywords: Gradient boosting; light GBM; machine learning; phishing; phishing URL; random forest

Paper 108: Ensemble of Deep Learning Models for Multi-plant Disease Classification in Smart Farming

Abstract: Plant disease identification at an early stage plays a crucial role in ensuring efficient management of the diseases and crop protection. The occurrence of plant ailments can result in substantial reductions in both crop yield and quality, which may cause financial setbacks for farmers and lead to food shortages for consumers. Traditional methods of disease detection rely on visual observation, which can consume a significant amount of time, be a labor-intensive, and often be inaccurate. Automated disease detection systems, based on techniques for machine learning have the potential to greatly improve the precision and speed of disease detection. This article presents a model for classifying plant diseases that combines the output of two transfer learning models, EfficientNetB0 and MobileNetV2, to improve disease classification accuracy. The PlantVillage Dataset was used to train and test the model under consideration, which contains 54,305 photos of 38 different plant disease classes, achieving an accuracy rate of 99.77% in disease classification. The use of an ensemble of deep learning models in this study shows promising results, indicating that the technique can enhance the accuracy of plant disease classification. Besides, this study contributes to the development of accurate and reliable automated disease detection systems, thereby supporting sustainable agriculture and global food security.

Author 1: Hoang-Tu Vo

Author 2: Luyl-Da Quach

Author 3: Hoang Tran Ngoc

Keywords: Ensemble learning; automated disease detection systems; transfer learning models; plant diseases

Paper 109: Primal-Optimal-Binding LPNet: Deep Learning Architecture to Predict Optimal Binding Constraints of a Linear Programming Problem

Abstract: Identifying an optimal basis for a linear programming problem is a challenging learning task. Traditionally, an optimal basis is obtained via the iterative simplex method which improves from the current basic feasible solution to the adjacent one until it reaches optimal. The obtained result is the value of the optimal solution and the corresponding optimal basis. Even though learning the optimal value is hard but learning the optimal basis is possible via deep learning. This paper presents the primal-optimal-binding LPNet that learns from massive linear programming problems of various sizes casting as all-unit-row-except-first-unit-column matrices. During the training step, these matrices are fed to the special row-column convolutional layer followed by the state-of-the-art deep learning architecture and sent to two fully connected layers. The result is the probability vector of non-negativity constraints and the original linear programming constraints at the optimal basis. The experiment shows that this LPNet achieves 99% accuracy of predicting a single binding optimal constraint on unseen test problems and Netlib problems. It identifies correctly 80% LP problems having all optimal binding constraints and faster than cplex solution time.

Author 1: Natdanai Kafakthong

Author 2: Krung Sinapiromsaran

Keywords: Deep learning; convolution neural network; linear programming; basic feasible solution; optimization

Paper 110: Detection of Epileptic Seizures Based-on Channel Fusion and Transformer Network in EEG Recordings

Abstract: According to the World Health Organization, epilepsy affects more than 50 million people in the world, and specifically, 80% of them live in developing countries. Therefore, epilepsy has become among the major public issue for many governments and deserves to be engaged. Epilepsy is characterized by uncontrollable seizures in the subject due to a sudden abnormal functionality of the brain. Recurrence of epilepsy attacks change people’s lives and interferes with their daily activities. Although epilepsy has no cure, it could be mitigated with an appropriated diagnosis and medication. Usually, epilepsy diagnosis is based on the analysis of an electroencephalogram (EEG) of the patient. However, the process of searching for seizure patterns in a multichannel EEG recording is a visual demanding and time consuming task, even for experienced neurologists. Despite the recent progress in automatic recognition of epilepsy, the multichannel nature of EEG recordings still challenges current methods. In this work, a new method to detect epilepsy in multichannel EEG recordings is proposed. First, the method uses convolutions to perform channel fusion, and next, a self-attention network extracts temporal features to classify between interictal and ictal epilepsy states. The method was validated in the public CHB-MIT dataset using the k-fold cross-validation and achieved 99.74% of specificity and 99.15% of sensitivity, surpassing current approaches.

Author 1: Jose Yauri

Author 2: Manuel Lagos

Author 3: Hugo Vega-Huerta

Author 4: Percy De-La-Cruz-VdV

Author 5: Gisella Luisa Elena Maquen-Ni˜no

Author 6: Enrique Condor-Tinoco

Keywords: Epilepsy; epilepsy detection; EEG; EEG channel fusion; convolutional neural network; self-attention

Paper 111: Light Field Spatial Super-resolution via Multi-level Perception and View Reorganization

Abstract: Light field (LF) imaging can obtain spatial and angular information of three-dimensional (3D) scene through a single shot, which enables a wide range of applications in the fields of 3D reconstruction, refocusing, virtual reality, etc. However, due to the inherent trade-off problem, the spatial resolution of acquired LF images is low, which hinders the widespread application of LF imaging technique. In order to relieve this issue, an end-to-end LF spatial super-resolution network is proposed by considering the multi-level perception and view reorganization. This method can fully explore the highly interwoven LF spatial and angular structure information. Specifically, a multi-feature fusion enhancement block is introduced that can fully perceive LF spatial, angular, and EPI information for LF spatial super-resolution. Furthermore, the angular coherence between LF views is exploited by reorganizing the LF sub-aperture images and constructing a multi-angular stack structure. Compared with other state-of-the-art methods, the proposed method achieves superior performance in both visual and quantitative terms.

Author 1: Yifan Mao

Author 2: Zaidong Tong

Author 3: Xin Zheng

Author 4: Xiaofei Zhou

Author 5: Youzhi Zhang

Author 6: Deyang Liu

Keywords: Light field image; spatial super-resolution; multi-level perception; view reorganization

Paper 112: Enhancing Intrusion Detection Systems with XGBoost Feature Selection and Deep Learning Approaches

Abstract: As cyber-attacks evolve in complexity and frequency; the development of effective network intrusion detection systems (NIDS) has become increasingly important. This paper investigates the efficacy of the XGBoost algorithm for feature selection combined with deep learning (DL) techniques, such as ANN, 1DCNN, and BiLSTM, to create accurate intrusion detection systems (IDSs) and evaluating it against NSL-KDD, CIC-IDS2017, and UNSW-NB15 datasets. The high accuracy and low error rate of the classification models demonstrate the potential of the proposed approach in IDS design. The study applied the XGBoost feature extraction technique to obtain a reduced feature vector and addressed data imbalance using the synthetic minority oversampling technique (SMOTE), signif-icantly improving the models’ performance in terms of precision and recall for individual attack classes. The ANN + BiLSTM model combined with SMOTE consistently out performed other models within this paper, emphasizing the importance of data balancing techniques and the effectiveness of integrating XGBoost and DL approaches for accurate IDSs. Future research can focus on implementing novel sampling techniques explicitly designed for IDSs to enhance minority class representation in public datasets during training.

Author 1: Khalid A. Binsaeed

Author 2: Alaaeldin M. Hafez

Keywords: Intrusion detection system; deep learning (DL); XG-Boost; feature extraction; Bidirectional Long Short-Term Memory (BiLSTM); Artificial Neural Networks (ANN); 1D Convolutional Neural Network (1DCNN); Synthetic Minority Oversampling Tech-nique (SMOTE); NSL-KDD dataset; CIC-IDS2017; UNSW-NB15

Paper 113: QMX-BdSL49: An Efficient Recognition Approach for Bengali Sign Language with Quantize Modified Xception

Abstract: Sign language is developed to bridge the com-munication gap between individuals with and without hearing impairment or speech difficulties. Individuals with hearing and speech impairment typically rely on hand signs as a means of expressing themselves. However, people, in general, may not have sufficient knowledge of sign language, thus a sign language recognition system on an embedded device is most needed. Literature related to such systems on embedded devices is scarce as these recognition tasks are very complex and computationally expensive. The limited resources of embedded devices cannot execute complex algorithms like Convolutional Neural Network (CNN) properly. Therefore, in this paper, we propose a novel deep learning architecture based on default Xception architec-ture, named Quantized Modified Xception (QMX) to reduce the model’s size and enhance the computational speed without compromising model accuracy. Moreover, the proposed QMX model is highly optimized due to the weight compression of model quantization. As a result, the footprint of the proposed QMX model is 11 times smaller than the Modified Xception (MX) model. To train the model, BDSL 49 dataset is utilized which includes approximately 14,700 images divided into 49 classes. The proposed QMX model achieves an overall F1 accuracy of 98%. In addition, a comprehensive analysis among QMX, Modified Xception Tiny (MXT), MX, and the default Xception model is provided in this research. Finally, the model has been implemented on Raspberry Pi 4 and a detailed evaluation of its performance has been conducted, including a comparison with existing state-of-the-art approaches in this domain. The results demonstrate that the proposed QMX model outperforms the prior work in terms of performance.

Author 1: Nasima Begum

Author 2: Saqib Sizan Khan

Author 3: Rashik Rahman

Author 4: Ashraful Haque

Author 5: Nipa Khatun

Author 6: Nusrat Jahan

Author 7: Tanjina Helaly

Keywords: Bengali sign language; CNN; computer vision; model quantization; Raspberry Pi 4; transfer learning; Tiny ML

Paper 114: Exploring Forest Transformation by Analyzing Spatial-temporal Attributes of Vegetation using Vegetation Indices

Abstract: The world’s ecosystem and environment are rapidly deteriorating with an increase in the depletion of forest conditions due to forest fires. In recent past years, wildfire incidents in Sikkim have increased due to severe climatic changes such as turbulent rainfall, untimely summers, extreme droughts in winters, and a reduction in the percentage of yearly rainfall. Forest fires are one of the numerous kinds of disasters that impose disastrous changes on the entire environment and disrupt the complex correspondence of the flora and fauna. The research’s goal is to examine the vegetation indices based on different climates to know why forest vegetation is decreasing day by day from 2000 to 2023. The frequent changes in forest vegetation are extensively studied by using satellite images. This data has been collected by three satellites Landsat-5, Landsat-8, and Landsat-9 on different vegetation indices NDVI, EVI, and NDWI. East Sikkim area is chosen to compute forest vegetation indices based on the heap’s landmass this region is unexplored yet and also studied about the forest changes by using different spatial temporal indices in the range of the entire district in the future. The authors of this paper have used Landsat multi-spectral data to assess changes in the area of vegetation in a sub-tropical region like a dense forest region in east Sikkim. The analysis depicts space images, computes vegetation indices (NDVI, EVI, NDWI), and accomplishes mathematical computation of findings. The proposed method will be helpful to discuss the variance of vegetation in the entire East Sikkim region at the time span of 2000–2023. In the analysis, we find that mean and standard deviation values change over the years in all indices. Later, we also calculated changes by using a classification model and find a total 10% change in forest areas in approximately 22 years.

Author 1: Anubhava Srivastava

Author 2: Sandhya Umrao

Author 3: Susham Biswas

Keywords: Classification; change detection; vegetation indices; landsat; machine learning

Paper 115: A Novel Mango Grading System Based on Image Processing and Machine Learning Methods

Abstract: Mangoes are a great commercial fruit and are widely cultivated in tropical areas. In smart agriculture, the automatic quality inspection and grading application is essential to post-harvest processing, due to the laborious nature and inconsistencies of traditional manual visual grading. This paper presents a low-cost, efficient, and effective mango grading system based on image processing and machine learning methods to generate higher quality fruit sorting, quality maintenance, production, and cut back labor concentration. A novel database of classified mangoes was collected and built in An Giang province. Methodologies and algorithms that utilize digital image processing, content-predicated analysis, and statistical analysis are implemented to determine the grade of local mango production. On our collected dataset, the proposed system achieved overall with an overall accuracy of 88% for all mango grades. The system shows compromised results for higher-quality fruit sorting, quality maintenance, and production while reducing labor concentration.

Author 1: Thanh-Nghi Doan

Author 2: Duc-Ngoc Le-Thi

Keywords: Smart agriculture; mango grading; image processing; machine learning methods

Paper 116: Mobile Module in Reconfigurable Intelligent Space: Applications and a Review of Developed Versions

Abstract: Due to the immobility of devices in conventional intelligent spaces, the quality and quantity of their applications (i.e., services) are thus restricted. To provide better and more applications, the devices in the spaces must be able to move autonomously to ideal positions. To solve this issue, the concepts of reconfigurable intelligent space (R+iSpace) and mobile modules (MoMos) have been introduced. Each device in the R+iSpace is carried by one or more MoMos that can freely move on the ceiling and walls. Consequently, the R+iSpace has evolved into a user-centered intelligent space, where devices can move to the user to provide services instead of the user having to move to where the devices are. In this work, several promising applications are introduced as open research challenges for the R+iSpace and the MoMo. In fact, various wall-climbing robots have been developed, however, their speed and carrying capacity are insufficient for adoption for the MoMo and the R+iSpace. Therefore, the development of MoMo requires the creation of entirely new designs and mechanisms. In addition to introducing promising applications, this work provides an overview of all versions of the MoMo that have been developed to gradually make it deployable in a realistic R+iSpace.

Author 1: Dinh Tuan Tran

Author 2: Tatsuki Satooka

Author 3: Joo-Ho Lee

Keywords: Climbing Robot; intelligent space; iSpace; mobile module; MoMo; reconfigurable intelligent space; R+iSpace; smart home; ubiquitous environment

Paper 117: A Fast and Accurate Deep Learning based Approach to Measure Fetal Fat from MRI Scan Images

Abstract: Retracted: After careful and considered review of the content of this paper by a duly constituted expert committee, this paper has been found to be in violation of IJACSA`s Publication Principles. We hereby retract the content of this paper. Reasonable effort should be made to remove all past references to this paper.

Author 1: Nagabotu Vimala

Author 2: Anupama Namburu

Keywords: Fetal adipose tissue; fetal MRI; automatic segmentation; radiologist

Paper 118: Blockchain-enabled Secure Privacy-preserving System for Public Health-center Data

Abstract: Health center data implicates a large scale of individual health records and is immensely concealment sensory. In the virtual era of large-size data, the increasingly different health informatization causes it important that health data needs to be stored precisely and securely. However, daily health data transactions carry the risk of privacy leaks that make sharing difficult. Moreover, the recently permitted blockchain applications suffer from deficient performance and lack of privacy. This study presents a privacy-preserving and secure sharing and storage system for public health centers based on the blockchain method to dispose of these issues. This system utilizes a hash-256-based access controller and transaction signature with the consensus policy and provides security to share and store health data in the blockchain. In this approach, blockchain guarantees scalability, privacy, integrity, and availability for data retention. Also, this paper measures the performance of transactions with supporting confidentiality-preserving and shows the average transaction time and acceptable latency when accessing health data.

Author 1: Md. Shohidul Islam

Author 2: Mohamed Ariff Bin Ameedeen

Author 3: Husnul Ajra

Author 4: Zahian Binti Ismail

Keywords: Blockchain; data; health; public; secure transaction

Paper 119: Video-based Heart Rate Estimation using Embedded Architectures

Abstract: Monitoring a driver’s heart rate is an important determinant to his health condition. The monitoring system must be accurate and non restrictive to the user’s actions. Estimating the driver’s change in his usual heart beat pattern can prevent undesirable outcomes. Several methods exist to estimate heart rate without any contact. In this paper, we are focusing on a method that uses remote photoplethysmography (rPPG). rPPG is a technique where heart rate is extracted from a PPG signal. The signal is extracted from the changes in blood flow that corresponds to the color variations recorded through an RGB camera. In this work, a different study that was based on an existing algorithm is presented to determine its processing time. The algorithm we proposed was divided into different global blocks and each block into different functional blocks (FBs). Though evaluating all the blocks’ processing time, it was possible to determine the most time consuming functional blocks. The results are implemented on different architectures: Desktop, Odroid XU4 and Jetson Nano to provide a higher performance.

Author 1: Hoda El Boussaki

Author 2: Rachid Latif

Author 3: Amine Saddik

Keywords: Heart rate; driver; photoplethysmography; non-contact; embedded architectures

Paper 120: Prediction of Death Counts Based on Short-term Mortality Fluctuations Data Series using Multi-output Regression Models

Abstract: Effective public health responses to unexpected epidemiological hazards or disasters need rapid and reliable monitoring. But, monitoring fast-changing situations and ac-quiring timely, accurate, and cross-national statistics to address short-term mortality fluctuations due to these hazards is very challenging. Estimating weekly excess deaths is the most solid and accurate way to measure the mortality burden caused by short-term risk factors. The Short-term Mortality Fluctuations (STMF) data series is one of the significant collections of the Human Mortality Database (HMD) that provides the weekly death counts and rates by age and sex of a country. Sometimes, the data collected from the sources are not always represented in specific age groups rather represented by the the total number of individual death records per week. However, the researchers reclassified their dataset based on the ranges of age and sex distributions of every country so that one can easily find out how many people died in per week of each country based on an equation and earlier distribution data. The paper focuses on the implementation of multi-output regression models such as logistic regression, decision tree, random forest, k nearest neighbors, lasso, support vector regressor, artificial neural network, and recurrent neural network to correctly predict death counts for specific age groups. According to the results, random forest delivered the highest performance with an R squared coefficient value of 0.9975, root mean square error of 43.2263, and mean absolute error of 16.4069.

Author 1: Md Imtiaz Ahmed

Author 2: Nurjahan

Author 3: Md. Mahbub-Or-Rashid

Author 4: Farhana Islam

Keywords: Multi-output regression model; short-term mortality fluctuations; machine learning; deep learning

Paper 121: Opportunities in Real Time Fraud Detection: An Explainable Artificial Intelligence (XAI) Research Agenda

Abstract: Regulatory and technological changes have recently transformed the digital footprint of credit card transactions, providing at least ten times the amount of data available for fraud detection practices that were previously available for analysis. This newly enhanced dataset challenges the scalability of traditional rule-based fraud detection methods and creates an opportunity for wider adoption of artificial intelligence (AI) techniques. However, the opacity of AI models, combined with the high stakes involved in the finance industry, means practitioners have been slow to adapt. In response, this paper argues for more researchers to engage with investigations into the use of Explainable Artificial Intelligence (XAI) techniques for credit card fraud detection. Firstly, it sheds light on recent regulatory changes which are pivotal in driving the adoption of new machine learning (ML) techniques. Secondly, it examines the operating environment for credit card transactions, an understanding of which is crucial for the ability to operationalise solutions. Finally, it proposes a research agenda comprised of four key areas of investigation for XAI, arguing that further work would contribute towards a step-change in fraud detection practices.

Author 1: Eleanor Mill

Author 2: Wolfgang Garn

Author 3: Nick Ryman-Tubb

Author 4: Chris Turner

Keywords: Artificial intelligence; explainable AI; machine learning; credit card fraud

Paper 122: Detecting Pneumonia with a Deep Learning Model and Random Data Augmentation Techniques

Abstract: This research paper presents an investigation into the detection of pneumonia using deep learning models and data augmentation techniques. The study compares and evaluates the performance of different models based on experimental results. The proposed model consists of multiple convolutional layers and maxpooling layers. Extensive experiments were conducted on a dataset, and the results demonstrate the efficiency and accuracy of our approach. The findings highlight the potential of deep learning in pneumonia detection and contribute to the existing body of knowledge in this field. The implications of this research can have a significant impact on improving diagnostic accuracy and patient outcomes. Future research directions could explore further enhancements in the model architecture, investigate additional data augmentation techniques, and consider larger datasets for more comprehensive evaluations.

Author 1: Tawfik Guesmi

Keywords: Deep learning; pneumonia detection; convolutional neural network; random data augmentation

Paper 123: Cross-age Face Image Similarity Measurement Based on Deep Learning Algorithms

Abstract: In this study, a multi-feature fusion and decoupling solution based on the RNN is proposed from a discriminative perspective. This method can address the identity and age information extraction losses in cross-age face recognition. This method not only constrains the correlation between identity and age using correlation loss but also optimizes identity feature restoration using feature decoupling. The model was trained and simulated in CACD and CACD-VS datasets. The single-task learning model stabilized after 125 iterations of training, while the multi-task learning model reached a stable and convergent state after 75 iterations. In terms of performance analysis, the DE-RNN model had the highest recognition accuracy with a mAP of 92.4%. The Human Voting model had a value of 90.2%. The mAP of the Human Average model was 81.8%, whereas the mAP of the DAL model was the lowest at 78.1%. Experiments proved that the model constructed in this study has effective recognition and application value in the cross-age face recognition scenario.

Author 1: Jing Zhang

Author 2: Ningyu Hu

Keywords: Cross-age; image recognition; RNN; feature fusion; decoupling; loss function