IJACSA Volume 15 Issue 3 - thesai.org

IJACSA Volume 15 Issue 3

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: Network Intrusion Detection in Cloud Environments: A Comparative Analysis of Approaches

Abstract: This research study comprehensively analyzes network intrusion detection in cloud environments by examining several approaches. These approaches have been explored and compared to determine the optimal and appropriate choice based on specific conditions. This research study employs a qualitative approach, specifically conducting a thematic literature analysis from 2020 to 2024. The research material has been exclusively obtained via Google Scholar. The traditional approaches identified in this research include anomaly-based and signature-based detection, along with innovative technologies and methods such as user behavior monitoring and machine learning. The findings of these studies demonstrate the effectiveness of conventional methods in known threat detection. They also struggle to identify novel attacks and understand the need for hybrid approaches that integrate the strengths of both. In this research study, the authors have addressed challenges such as privacy compliance, performance scalability, and false positives, highlighting the importance of continuous monitoring, privacy-preserving technologies, and real-time threat intelligence integration. This study also highlights the importance of stakeholder buy-in and staff training for the successful implementation of a network intrusion detection system (NIDS), especially when determining the evolving nature of cyber threats. This study concludes by defining a balanced approach combining new and old methodologies to offer an effective defense against diverse cyber threats in cloud environments. The future scope of NIDS in cloud environments has also been discussed, including enhancing privacy compliance capabilities and integrating AI-driven anomaly detection to meet emerging threats and regulatory requirements.

Author 1: Sina Ahmadi

Keywords: Cloud networking; cloud security; firewall; intrusion detection; NIDS

Paper 2: Analysis of Gait Motion Sensor Mobile Authentication with Machine Learning

Abstract: In recent decades, mobile devices have evolved in potential and prevalence significantly while advancements in security have stagnated. As smartphones now hold unprecedented amounts of sensitive data, there is an increasing need to resolve this gap in security. To address this issue, researchers have experimented with biometric-based authentication methods to improve smartphone security. Following a comprehensive review, it was found that gait-based mobile authentication is under-researched compared to other behavioral biometrics. This study aims to contribute to the knowledge of biometric and gait-based authentication through the analysis of recent gait datasets and their potential with machine learning algorithms. Two recently published gait datasets were used with algorithms such as Random Forest, Decision Tree, and XGBoost to successfully differentiate users based on their respective walking features. Throughout this paper, the datasets, methodology, algorithms, experimental results, and goals for future work will be described.

Author 1: Sara Kokal

Author 2: Mounika Vanamala

Author 3: Rushit Dave

Keywords: Machine learning; machine learning algorithms; behavioral biometrics; gait dynamics; motion sensors

Paper 3: Privacy-Aware Decision Making: The Effect of Privacy Nudges on Privacy Awareness and the Monetary Assessment of Personal Information

Abstract: Nowadays, smartphones are equipped with various sensors collecting a huge amount of sensitive personal information about their users. However, for smartphone users, it remains hidden, and sensitive information is accessed by used applications and data requesters. Moreover, governmental institutions have no means to verify if applications requesting sensitive informa-tion are compliant with the General Data Protection Directive (GDPR), as it is infeasible to check the technical details and data requested by applications that are on the market. Thus, this research aims to shed light on the compliance analysis of applications with the GDPR. Therefore, a multidimensional analysis is applied to analyzing the permission requests of applications and empirically test if the information provided about potentially dangerous permissions influences the privacy awareness and their willingness to pay or sell personal data of users. The use case of Google Maps has been chosen to examine privacy awareness and the monetary assessment of data in a concrete scenario. The information about the multidimensional analysis of the permission requests of Google Maps and the privacy consent form is used to design privacy nudges to inform users about potentially harmful permission requests that are not in line with the GDPR. The privacy nudges are evaluated in two crowdsourcing experiments with overall 426 participants, showing that information about harmful data collection practices increases privacy awareness and also the willingness to pay for the protection of personal data.

Author 1: Vera Schmitt

Author 2: James Nicholson

Author 3: Sebastian Moller

Keywords: Privacy protection; privacy policy analysis; GDPR; willingness to pay; privacy awareness

Paper 4: User Experience and Behavioural Adaptation Based on Repeated Usage of Vehicle Automation: Online Survey

Abstract: For years, Level 2 vehicle automation systems (VAS) have been commercially available, yet the extent to which users comprehend their capabilities and limitations remains largely unclear. This study aimed to evaluate user knowledge regarding Level 2 VAS and explore the correlation between user experiences (UX), behavioural adaptations, trust, and acceptance. By using an online survey, we sought to deepen understanding of how UX, trust, and acceptance of Level 2 automated vehicles (AVs) evolve with prolonged use in urban traffic. The survey, comprising demographic data and knowledge inquiries (automated driving experience and timeframes, vehicle operation competency, driving skills over long-term use of automation, the learning process, automation-induced effects, trust in automation, and ADS researchers and manufacturers), was completed by various drivers (N=16). This investigation focused on users' long-term experiences with automation in urban traffic. Consequently, we offer user-centric transformative insights into users' experiences with driving automation in urban traffic settings. Results revealed that users’ knowledge of automation exhibits their learning patterns, trust and acceptance. Moreover, users’ attitudes trust, and acceptance varies across different user profiles. What we have also learned about UX and the changing nature of user behaviours towards automation is that, automated driving changes influence the safety and risk conditions in which users and AVs interact. These findings can inform the development of interaction design strategies and policy aimed at enhancing UX of AV users.

Author 1: Naomi Y. Mbelekani

Author 2: Klaus Bengler

Keywords: Automated vehicles; automation effects; user experience (UX); trust; acceptance; behavioural adaptations

Paper 5: Generative Adversarial Neural Networks for Realistic Stock Market Simulations

Abstract: Stock market simulations are widely used to create synthetic environments for testing trading strategies before deploying them to real-time markets. However, the weak realism often found in these simulations presents a significant challenge. Improving the quality of stock market simulations could be facilitated by the availability of rich and granular real Limit Order Books (LOB) data. Unfortunately, access to LOB data is typically very limited. To address this issue, a framework based on Generative Adversarial Networks (GAN) is proposed to generate synthetic realistic LOB data. This generated data can then be utilized for simulating downstream decision-making tasks, such as testing trading strategies, conducting stress tests, and performing prediction tasks. To effectively tackle challenges related to the temporal and local dependencies inherent in LOB structures and to generate highly realistic data, the framework relies on a specific data representation and preprocessing scheme, transformers, and conditional Wasserstein GAN with gradient penalty. The framework is trained using the FI-2010 benchmark dataset and an ablation study is conducted to demonstrate the importance of each component of the proposed framework. Moreover, qualitative and quantitative metrics are proposed to assess the quality of the generated data. Experimental results indicate that the framework outperforms existing benchmarks in simulating realistic market conditions, thus demonstrating its effectiveness in generating synthetic LOB data for diverse downstream tasks.

Author 1: Badre Labiad

Author 2: Abdelaziz Berrado

Author 3: Loubna Benabbou

Keywords: Limit order book simulations; transformers; wasserstein GAN with gradient penalty; FI-2010 benchmark dataset

Paper 6: Integrating Generative AI for Advancing Agile Software Development and Mitigating Project Management Challenges

Abstract: Agile software development emphasizes iterative progress, adaptability, and stakeholder collaboration. It champions flexible planning, continuous improvement, and rapid delivery, aiming to respond swiftly to change and deliver value efficiently. Integrating Generative Artificial Intelligence (AI) into Agile software development processes presents a promising avenue for overcoming project management challenges and enhancing the efficiency and effectiveness of software development endeavors. This paper explores the potential benefits of leveraging Generative AI in Agile methodologies, aiming to streamline development workflows, foster innovation, and mitigate common project management challenges. By harnessing the capabilities of Generative AI for tasks such as code generation, automated testing, and predictive analytics, Agile teams can augment their productivity, accelerate delivery cycles, and improve the quality of software products. Additionally, Generative AI offers opportunities for enhancing collaboration, facilitating decision-making, and addressing uncertainties inherent in Agile project management. Through an in-depth analysis of the integration of Generative AI within Agile frameworks, this paper provides insights into how organizations can harness the transformative potential of AI to advance Agile software development practices and navigate the complexities of modern software projects more effectively.

Author 1: Anas BAHI

Author 2: Jihane GHARIB

Author 3: Youssef GAHI

Keywords: Artificial intelligence; software engineering; Agile software development

Paper 7: Comparing Regression Models to Predict Property Crime in High-Risk Lima Districts

Abstract: Crime continues to be an issue, in Metropolitan Lima, Peru affecting society. Our focus is on property crimes. We recognized the lack of studies on predicting these crimes. To tackle this problem, we used regression techniques such as XGBoost, Extra Tree, Support Vector, Bagging, Random Forest and AdaBoost. Through GridsearchCV we optimized hyperparameters to enhance our research findings. The results showed that Extra Tree Regression stood out as the model with an R2 value of 0.79. Additionally, error metrics like MSE (185.43) RMSE (13.62) and MAE (10.47) were considered to evaluate the model's performance. Our approach considers time patterns in crime incidents. Contributes, to addressing the issue of insecurity in a meaningful way.

Author 1: Maria Escobedo

Author 2: Cynthia Tapia

Author 3: Juan Gutierrez

Author 4: Victor Ayma

Keywords: Supervised techniques; machine learning; regression; crime; prediction

Paper 8: Educational Performance Prediction with Random Forest and Innovative Optimizers: A Data Mining Approach

Abstract: In the ever-evolving landscape of education, institutions grapple with the intricate task of evaluating individual capabilities and forecasting student performance. Providing timely guidance becomes pivotal, steering students toward specific areas for focused academic enhancement. Within the educational domain, the utilization of data mining emerges as a powerful tool, revealing latent patterns within vast datasets. This study adopts the Random Forest classifier (RFC) for predicting student performance, bolstered by the integration of two innovative optimizers—Victoria Amazonia Optimization (VAO) and Phasor Particle Swarm Optimizer (PPSO). A notable contribution of this research lies in the introduction of these novel optimizers to augment the model's accuracy, elevating the precision of predictions. Robust evaluation metrics, including Accuracy, Precision, Recall, and F1-score, meticulously gauge the model's effectiveness in this context. Remarkably, the results underscore the supremacy of RFC+VAO, showcasing exceptional values for Accuracy (0.934), Precision (0.940), Recall (0.930), and F1-score (0.930). This substantiates the significant contribution of integrating VAO into the Random Forest framework, promising substantial advancements in predictive analytics for educational institutions. The findings not only accentuate the efficacy of the proposed methodology but also herald a new era of precision and reliability in predicting student performance, thereby enriching the landscape of educational data analytics.

Author 1: Yanli Chen

Author 2: Ke Jin

Keywords: Student performance; Random Forest Classification; victoria amazonia; phasor particle swarm

Paper 9: Fuzzy Deep Learning Approach for the Early Detection of Degenerative Disease

Abstract: Degenerative diseases can impact individuals of any age, encompassing children and teenagers; however, they typically tend to affect productive or adult individuals. Globally, conventional and advanced diagnostic methods, including those developed in Indonesia, have emerged to identify and manage these health conditions. Problems in brain tumor detection are the intricate process of precisely and effectively identifying the presence of tumors in the brain. On the other hand, diagnosing brain tumors in the laboratory poses issues related to time consumption, inaccuracy, lack of consistency, and costliness. This study specifically concentrates on the early detection of brain tumors by analyzing images generated through MRI scans. Unlike the traditional method of manual image analysis conducted by seasoned physicians, our approach integrates fuzzy logic to enable the early identification of brain tumors. The principal objective of this research is to enhance understanding and develop an intelligent, swift, and precise application for diagnosing brain tumors using medical imaging. The segmentation technique provides practical technology for the early detection of brain tumors. Utilizing a dataset comprising over 13,000 data points and undergoing a year-long training process with approximately 1,310 MRI images, the research culminates in the creation of a tool or software application system for the analysis of medical images. Despite the impressive precision score of 0.9992, highlighting its exceptional accuracy in correctly identifying positive instances, the recall value of 0.5767 suggests the potential exclusion of a significant number of actual positive instances in its predictions.

Author 1: Chairani

Author 2: Suhendro Y. Irianto

Author 3: Sri Karnila

Author 4: Adimas

Keywords: Degenerative diseases; brain tumor; fuzzy; deep learning

Paper 10: Offline Author Identification using Non-Congruent Handwriting Data Based on Deep Convolutional Neural Network

Abstract: This investigation presents a novel technique for offline author identification using handwriting samples across diverse experimental conditions, addressing the intricacies of various writing styles and the imperative for organizations to authenticate authorship. Notably, the study leverages inconsistent data and develops a method independent of language constraints. Utilizing a comprehensive dataset adhering to American Society for Testing and Materials (ASTM) standards, a deep convolutional neural network (DCNN) model, enhanced with pre-trained networks, extracts features hierarchically from raw manuscript data. The inclusion of heterogeneous data underscores a significant advantage of this study, while the applicability of the proposed DCNN model to multiple languages further highlights its versatility. Experimental results demonstrate the efficacy of the proposed method in author identification. Specifically, the proposed model outperforms conventional approaches across four comprehensive datasets, exhibiting superior accuracy. Comparative analysis with engineering features and traditional methods such as Support Vector Machine (SVM) and Backpropagation Neural Network (BPNN) underscores the superiority of the proposed technique, yielding approximately a 13% increase in identification accuracy while reducing reliance on expert knowledge. The validation results, showcase the diminishing network error and increasing accuracy, with the proposed model achieving 99% accuracy after 200 iterations, surpassing the performance of the LeNet model. These findings underscore the robustness and utility of the proposed technique in diverse applications, positioning it as a valuable asset for handwriting recognition experts.

Author 1: Ying LIU

Author 2: Gege Meng

Author 3: Naiyue ZHANG

Keywords: Handwriting recognition; offline author identification; deep convolutional neural network; image processing; language versatility; feature extraction; hierarchical model

Paper 11: An Efficient and Intelligent System for Controlling the Speed of Vehicle using Fuzzy Logic and Deep Learning

Abstract: Vehicle collisions are a significant problem worldwide, causing injuries, fatalities, and property damage. There are several reasons for the collapse of vehicles such as rash driving, over speeding, less driving skills, increasing number of vehicles, drunk and drive, etc. However, over speeding is one of the critical factors out of all the reasons for vehicle collisions. To address the critical issues, the current article proposes a Fuzzy-based algorithm to prevent and control the speed of the vehicle. The major objective of the proposed system is to control the speed of the vehicle for proactive collision avoidance. Deep learning and fuzzy system provide better integrated approach for the controlling of the speed and avoid vehicle collision. Fuzzification of the speed variable provides an advanced or viable solution for speed control. The current research used RNN and other deep learning algorithm to predict the traffic and identify the traffic frequency. The traffic frequency in a time-series frame provides the frequency of the traffic within a time frame that can be detected by using involvement of IoT.

Author 1: Anup Lal Yadav

Author 2: Sandip Kumar Goyal

Keywords: Speed control; fuzzy logics; deep learning; decision making; collision avoidance

Paper 12: A Single Stage Detector for Breast Cancer Detection on Digital Mammogram

Abstract: Medical image processing plays a pivotal role in modern healthcare, and the early detection of breast cancer in digital mammograms. Several methods have been explored in the literature to improve breast cancer detection, with deep-learning approaches emerging as particularly promising due to their ability to provide accurate results. However, a persistent research challenge in deep learning-based breast cancer detection lies in addressing the historically low accuracy rates observed in previous studies. This paper presents a novel deep-learning model utilizing a single-stage detector based on the YOLOv5 algorithm, designed specifically to tackle the issue of low accuracy in breast cancer detection. The proposed method involves the generation of a custom dataset and subsequent training, validation, and testing phases to evaluate the model's performance rigorously. Experimental results and comprehensive performance evaluations demonstrate that the proposed method achieves remarkable accuracy, marking a significant advancement in breast cancer detection through extensive experiments and rigorous performance analysis.

Author 1: Li Xu

Author 2: Nan Jia

Author 3: Mingmin Zhang

Keywords: Breast cancer detection; digital mammogram; deep learning; YOLOv5 algorithm; medical image processing

Paper 13: Introducing an Innovative Approach to Mitigate Investment Risk in Financial Markets: A Case Study of Nikkei 225

Abstract: When the value of an investor's stock portfolio rises during a period of great market performance, investors often experience a gain in wealth. Spending may increase when people feel more at ease and confident about their financial circumstances. On the other hand, during a market crisis, a fall in wealth could lead to lower consumer spending, which could impede economic growth. Stock market trend prediction is thought to be a more important and fruitful endeavor. Stock prices will, therefore, provide significant returns from prudent investing decisions. Because of the outdated and erratic data, stock market forecasts pose a serious challenge to investors. As a result, stock market forecasting is among the main challenges faced by investors trying to optimize their return on investment. The goal of this research is to provide an accurate hybrid stock price forecasting model using Nikkei 225 index data from 2013 to 2022. The construction of the support vector regression involves the integration of multiple optimization approaches, including moth flame optimization, artificial bee colony, and genetic algorithms. Moth flame optimization is proven to produce the best results out of all of these optimization techniques. The evaluation criteria used in this research are MAE, MAPE, MSE, and RMSE. The results obtained for MFO-SVR, which is 0.70 for criterion MAPE, show the high accuracy of this model for estimating the price of Nikkei 225.

Author 1: Xiao Duan

Keywords: NIKKEI 225 index; artificial bee colony; stock price; financial markets; support vector regression

Paper 14: Clustering Algorithms in Sentiment Analysis Techniques in Social Media – A Rapid Literature Review

Abstract: Based on the high dynamic of Sentiment Analysis (SA) topic among the latest publication landscape, the current review attempts to fill a research gap. Consequently, the paper elaborates on the most recent body of literature to extract and analyze the papers that elaborate on the clustering algorithms applied on social media datasets for performing SA. The current rapid review attempts to answer the research questions by analyzing a pool of 46 articles published in between Dec 2020 – Dec 2023. The manuscripts were thoroughly selected from Scopus (Sco) and WebOf-Science (WoS) databases and, after filtering the initial pool of 164 articles, the final results (46) were extracted and read in full.

Author 1: Vasile Daniel Pavaloaia

Keywords: Clustering algorithms; K-means; HAC; DBSCAN; sentiment analysis; natural language processing techniques; social media datasets; Twitter/X

Paper 15: A Systematic Review of the Literature on the Use of Artificial Intelligence in Forecasting the Demand for Products and Services in Various Sectors

Abstract: This systematic review, carried out under the PRISMA methodology, aims to identify the recently proposed artificial intelligence models for demand forecasting, distinguishing the problems they try to overcome, recognizing the artificial intelligence methods used, detailing the performance metrics used, recognizing the performance achieved by these models and identifying what is new in them. Studies in the manufacturing, retail trade, tourism and electric energy sectors were considered in order to facilitate the transfer of knowledge from different sectors. 33 articles were analyzed, with the main results being that the proposed models are generally ensembles of various artificial intelligence methods; that the complexity of data and its scarcity are the main problems addressed; that combinations of simple machine learning, “bagging”, “boosting” and deep neural networks, are the most used methods; that the performance of the proposed models surpasses the classic statistical methods and other reference models; and that, finally, the proposed novelties cover aspects such as the type of data used, the pattern extraction techniques used, the assembly forms of the applied models and the use of algorithms for automating the adjustment of the models. Finally, a forecast model is proposed that includes the most innovative aspects found in this research.

Author 1: José Rolando Neira Villar

Author 2: Miguel Angel Cano Lengua

Keywords: Demand; agglomeration algorithm; services; PRISMA methodology; artificial intelligence

Paper 16: Integration of Effective Models to Provide a Novel Method to Identify the Future Trend of Nikkei 225 Stocks Index

Abstract: The stock market refers to a financial market in which individuals and institutions engage in the buying and selling of shares of publicly listed firms. The valuation of stocks is influenced by the interplay between the forces of supply and demand. The act of allocating funds to the stock market entails a certain degree of risk, while it presents the possibility of substantial gains over an extended period. The task of predicting stock prices in the securities market is further complicated by the presence of non-stationary and non-linear characteristics in financial time series data. While traditional techniques have the potential to enhance the accuracy of forecasting, they are also associated with computational complexities that might lead to an elevated occurrence of prediction mistakes. This is the reason why the financial industry has seen a growing prevalence of novel methods, particularly in the stock market. This work introduces a novel model that effectively addresses several challenges by integrating the random forest methodology with the artificial bee colony algorithm. In the current study, the hybrid model demonstrated superior performance and effectiveness compared to the other models. The proposed model exhibited optimum performance and demonstrated a significant degree of effectiveness with low errors. The efficiency of the predictive model for stock price forecasts was established via the analysis of data obtained from the Nikkei 225 index. The data included the timeframe from January 2013 to December 2022. The results reveal that the proposed framework demonstrates efficacy and reliability in evaluating and predicting the price time series of equities. The empirical evidence suggests that, when compared to other current methodologies, the proposed model has a greater degree of accuracy in predicting outcomes.

Author 1: Jiang Zhu

Author 2: Haiyan Wu

Keywords: Financial market; stock future trend; Nikkei 225 index; random forest; artificial bee colony

Paper 17: A CNN-based Deep Learning Framework for Driver’s Drowsiness Detection

Abstract: Accidents are one of the major causes of injuries and deaths worldwide. According to the WHO report, in 2022 an estimated 1.3 million people die from road accidents. Driver fatigue is the primary factor in these traffic accidents. There are a number of studies presented by previous researchers in the context of driver’s drowsiness detection. The majority of earlier strategies relied on image processing systems that used algorithms to identify the yawning, eye closure, and eyebrow of the driver taken from the live video camera. One of the major issues of the previous studies was the delay in detection time and dataset. These studies used physical sensors for monitoring the driver’s behavior causes in delay time of detection. In this article, a deep learning approach is used to provide a continuous strategy for detecting driver’s drowsiness using an efficient dataset. The trained algorithm is employed on the video taken from the live camera to extract the driver's facial landmarks, which are subsequently processed by a trained algorithm to provide results. The dataset used for training the CNN algorithm is consisting of 2904 images taken from various subjects under various driving circumstances. The data is preprocessed by different methods including statistical moments, CNN filters, frequency vector determination and position Incidence vector calculation. After training the algorithm the feature-based cascade classifiers files are used to recognize the face from the real-life scenario using the live camera. The accuracy of the purposed model is 95%, which is the highest of all the purposed models, based on data gathered from different kind of scenarios.

Author 1: Ali Sohail

Author 2: Asghar Ali Shah

Author 3: Sheeba Ilyas

Author 4: Nizal Alshammry

Keywords: Drowsiness detection face detection; eye detection; yawn detection; deep learning; convolutional neural network; electroencephalograph; eye aspect ratio

Paper 18: A Fire and Smoke Detection Model Based on YOLOv8 Improvement

Abstract: The warning of fire and smoke provides security for people's lives and properties. The utilization of deep learning for fire and smoke warning has been an active area of research, especially the use of target detection algorithms has achieved significant results. For improving the fire and smoke detection performance of model in different scenarios, a high-precision and lightweight improvement based on the model of You Only Look Once (YOLO), is developed. It utilizes partial convolutions to reduce the complexity of model, and add an attention block to acquire the cross-space learning capability. In addition, the neck network is redesigned to realize bidirectional feature fusion. Experiments show that it has significantly improved the results for all metrics in the public Fire-Smoke dataset, and the size of the model has also been widely reduced. Comparisons with other popular target detection models under the same conditions indicate that the improved model has the best performance as well. In order to have a more visual comparison with the detectability of the original model, the heatmap experiments are also established, which also demonstrate that it is characterized by less leakage rate and more focused attention.

Author 1: Pengcheng Gao

Keywords: Fire and smoke detection; deep learning; computer vision; YOLO

Paper 19: Prediction of Cardiovascular Disease using Machine Learning Algorithms

Abstract: Heart is the most critical organ of our body for being responsible for regulating and maintaining the blood circulation levels. Globally, heart disease cases are prevalent and constitute a significant cause of mortality. Manifestations such as chest discomfort and irregular heartbeat are notable symptoms. The healthcare sector has amassed substantial knowledge in this domain. Analyzing the research, this paper delves into the concept of utilizing ML algorithms to predict cardiac diseases. In this research will employ a diverse array of machine learning techniques, including decision tree, support vector classifier, random forest, K-NN, logistic regression and naive Bayes. These algorithms utilize specific characteristics to forecast cardiac diseases effectively. Leveraging machine learning algorithms to analyze and predict outcomes from the extensive healthcare-generated data shows considerable promise. Recent advancements in machine learning models have incorporated numerous features, and in this study, propose the integration of these features in machine learning algorithms to forecast cardiovascular ailments. The main objective of this research is to identify the performance of the mentioned machine learning algorithms for predicting cardiovascular elements.

Author 1: Mahesh Kumar Joshi

Author 2: Deepak Dembla

Author 3: Suman Bhatia

Keywords: Cardiovascular disease; heart; logistic regression; K-NN; machine learning; naïve bayes; SVM

Paper 20: Underwater Image Enhancement via Higher-Order Moment CLAHE Model and V Channel Substitute

Abstract: Images captured underwater often exhibit low contrast and color distortion attributed to special properties of light in water. Underwater image enhancement methods have become an effective solution to address these issues due to its simplicity and effectiveness. However, underwater image enhancement methods (such as CLAHE) face challenge of increasing image contrast, improve generalization of method. Here, underwater image enhancement via higher-order moment CLAHE model and V channel substitute is proposed to enhance contrast and correct color distortion. Firstly, analyze statistical features of image histograms, use higher-order moments to quantify features in a targeted manner, add them to CLAHE, so that improved CLAHE can accurately enhance contrast of underwater image according to dynamic features of image blocks, avoiding over- or under-enhancement of image. Then, for problem of color distortion, this paper novelty uses gray data to substitutes V channel in HSV color space, compensated for lost information, so as to achieve purpose of color correction in terms of visual perception. Finally, color correction of image through gray world method, which effectively improve color distortion problem. Our method is qualitatively and quantitatively compared with multiple state-of-the-art methods in public dataset, demonstrating that this method better solved low contrast and color distortion, in addition, details were more realistic, and evaluation indexes of underwater image quality were better.

Author 1: Chen Yahui

Author 2: Liang Yitao

Author 3: Li Yongfeng

Author 4: Liu Hongyue

Author 5: Li Lan

Keywords: Underwater images; contrast enhancement; adaptive CLAHE; high-order moments; dynamic features

Paper 21: Towards Digital Preservation of Cultural Heritage: Exploring Serious Games for Songket Tradition

Abstract: Over the past few decades, Malaysia has undergone remarkable technological advancement, establishing itself as a vibrant hub for innovation in Southeast Asia. However, technological progress must be harmonized with preserving and promoting the country's cultural heritage. Digital preservation of cultural heritage emerges as a critical endeavor, particularly for future generations. Nonetheless, there remains a notable deficiency in preservation methodologies for cultural heritage, particularly concerning technological approaches. This paper delves into the realm of cultural heritage and presents findings from a study on preserving Songket's heritage. Interviews were conducted with three experts on Songket heritage, revealing a prevailing lack of awareness regarding Songket heritage preservation. Additionally, the analysis highlights inherent flaws in current preservation methods, hindering efforts to engage a wider audience, particularly the younger generation. The experts unanimously advocate digitizing heritage knowledge, including the integration of serious games, to facilitate Songket preservation and safeguarding efforts. The use of serious games can also attract and engage the younger generation to the heritage of Songket.

Author 1: Nor Hafidzah Abdullah

Author 2: Wan Malini Wan Isa

Author 3: Syadiah Nor Wan Shamsuddin

Author 4: Norkhairani Abdul Rawi

Author 5: Maizan Mat Amin

Author 6: Wan Mohd Adzim Wan Mohd Zain

Keywords: Cultural Heritage; digital preservation; serious game; Songket

Paper 22: RUICP: Commodity Recommendation Model Based on User Real Time Interest and Commodity Popularity

Abstract: At present, the recommendation of massive commodities mainly depends on the short-term click through rate of commodities and the data directly browsed and clicked by users. This recommendation method can better meet the shopping needs of users, but there are two shortcomings. One is to recommend homogeneous commodities to long-term shopping users; second, we can't grasp the real-time changes of users' interests, and can only recommend results similar to the recently clicked products. Therefore, this study intends to establish a time-varying expression method of users' interest intensity to solve the deviation of real-time recommendation content, and propose a recommendation model RUICP based on users' time-dependent interest and commodity heat. Firstly, the user's basic data and cumulative usage information are used for portrait, specifically, the user's usage data is divided into isochronous and deep-seated semantic feature analysis, the model is optimized and the user's long-term interest intensity is obtained after parameter estimation; Then, the user's short-term interest is obtained by splitting the user's short-term use data, and the user's final interest is calculated by combining the short-term interest and the user's long-term interest intensity; Then calculate the product popularity score by adding the repeated click through rate of products, and then update the ranking of products; Finally, the classic item based collaborative filtering algorithm is used to calculate the matching degree of user interest and goods, and then recommend. The results of simulation experiments show that compared with other methods, RUICP has higher recommendation accuracy for old users and has certain value for solving the cold start problem.

Author 1: Wenchao Xu

Author 2: Ling Xia

Keywords: User real time interest; commodity popularity; recommend

Paper 23: Beyond BERT: Exploring the Efficacy of RoBERTa and ALBERT in Supervised Multiclass Text Classification

Abstract: This study investigates the performance of transformer-based machine learning models, specifically BERT, RoBERTa, and ALBERT, in multiclass text classification within the context of the Universal Access to Quality Tertiary Education (UAQTE) program. The aim is to systematically categorize and analyze qualitative responses to uncover domain-specific patterns in students' experiences. Through rigorous evaluation of various hyperparameter configurations, consistent enhancements in model performance are observed with smaller batch sizes and increased epochs, while optimal learning rates further boost accuracy. However, achieving an optimal balance between sequence length and model efficacy presents nuanced challenges, with instances of overfitting emerging after a certain number of epochs. Notably, the findings underscore the effectiveness of the UAQTE program in addressing student needs, particularly evident in categories such as "Family Support" and "Financial Support," with RoBERTa emerging as a standout choice due to its stable performance during training. Future research should focus on fine-tuning hyperparameter values and adopting continuous monitoring mechanisms to reduce overfitting. Furthermore, ongoing review and modification of educational efforts, informed by evidence-based decision-making and stakeholder feedback, is critical to fulfill students' changing needs effectively.

Author 1: Christian Y. Sy

Author 2: Lany L. Maceda

Author 3: Mary Joy P. Canon

Author 4: Nancy M. Flores

Keywords: Multi-class text classification; Bidirectional Encoder Representations from Transformers (BERT); RoBERTa; ALBERT; Universal Access to Quality Tertiary Education (UAQTE) program; educational policy reforms

Paper 24: Assessment of Attention-based Deep Learning Architectures for Classifying EEG in ADHD and Typical Children

Abstract: Although limited research has explored the integration of electroencephalography (EEG) and deep learning approaches for attention deficit hyperactivity disorder (ADHD) detection, using deep learning models for actual data, including EEGs, remains a difficult endeavour. The purpose of this work was to evaluate how different attention processes affected the performance of well-established deep-learning models for the identification of ADHD. Two specific architectures, namely long short-term memory (LSTM)+ attention (Att) and convolutional neural network (CNN)s+Att, were compared. The CNN+Att model consists of a dropout, an LSTM layer, a dense layer, and a CNN layer merged with the convolutional block attention module (CBAM) structure. On top of the first LSTM layer, an extra LSTM layer, including T LSTM cells, was added for the LSTM+Att model. The information from this stacked LSTM structure was then passed to a dense layer, which, in turn, was connected to the classification layer, which comprised two neurons. Experimental results showed that the best classification result was achieved using the LSTM+Att model with 98.91% accuracy, 99.87% accuracy, 97.79% specificity and 98.87% F1-score. After that, the LSTM, CNN+Att, and CNN models succeeded in classifying ADHD and Normal EEG signals with 98.45%, 97.74% and 97.16% accuracy, respectively. The information in the data was successfully utilized by investigating the application of attention mechanisms and the precise position of the attention layer inside the deep learning model. This fascinating finding creates opportunities for more study on large-scale EEG datasets and more reliable information extraction from massive data sets, ultimately allowing links to be made between brain activity and specific behaviours or task execution.

Author 1: Mingzhu Han

Author 2: Guoqin Jin

Author 3: Wei Li

Keywords: ADHD; EEG; deep learning; attention mechanisms; CNN; LSTM

Paper 25: Precision Face Mask Detection in Crowded Environment using Machine Vision

Abstract: In the face of rampant global disease transmission, effective preventive strategies are imperative. This study tackles the challenge of ensuring compliance in crowded settings by developing a sophisticated face mask detection system. Utilizing MATLAB and the Cascade Object detector, the system focuses on detecting white surgical masks in frontal images. Training the system is critical for accuracy; therefore, cross-validation is employed due to limited data. The results reveal accuracies of 76.67% for initial training, 67.50% for a 9:11 cropping ratio, and 89.17% for a 9:4:7 cropping ratio, highlighting the system's remarkable precision in mask detection. Looking ahead, the system's adaptability can be further expanded to include various mask colors and types, extending its effectiveness beyond COVID-19 to combat a range of respiratory illnesses. This research represents a significant advancement in reinforcing preventive measures against future disease outbreaks, especially in densely populated environments, contributing significantly to global public health and safety initiatives.

Author 1: Jamil Abedalrahim Jamil Alsayaydeh

Author 2: Mohd Faizal bin Yusof

Author 3: Chan Yoke Lin

Author 4: Mohammed Nasser Mohammed Al-Andoli

Author 5: Safarudin Gazali Herawan

Author 6: Ida Syafiza Md Isa

Keywords: Face mask detection; machine vision; cascade object detector; cross-validation

Paper 26: Predicting Obesity in Nutritional Patients using Decision Tree Modeling

Abstract: Obesity has become a widespread problem that affects not only physical well-being but also mental health. To address this problem and provide solutions, Machine Learning (ML) technology tools are being applied. Studies are currently being developed to improve the prediction of obesity. This study aimed to predict obesity levels in nutritional patients by analyzing their physical and dietary habits using the Decision Tree (DT) model. For the development of this work, we chose to use the CRISP-DM framework to follow the development in an organized way, thus achieving a better understanding of the data and describing, evaluating, and analyzing the results. The results of this work yielded metrics with significant values for predicting obesity: so much so that the accuracy rate was 92.89%, the sensitivity rate was 94% and the F1 score was 93%. Likewise, accuracy metrics above 88% were obtained for each level of obesity, demonstrating the effectiveness of the DT model in predicting this type of task. Finally, the results demonstrate that the DT model is effective in predicting obesity, with significant results that motivate further research to continue improving accuracy in this type of task.

Author 1: Orlando Iparraguirre-Villanueva

Author 2: Luis Mirano-Portilla

Author 3: Manuel Gamarra-Mendoza

Author 4: Wilmer Robles-Espiritu

Keywords: Obesity; Machine Learning (ML); Decision Tree (DT); Prediction; CRISP-DM

Paper 27: Presenting a Hybrid Method to Overcome the Challenges of Determining the Uncertainty of Future Stock Price Identification

Abstract: A particular location, framework, or forum where buyers and sellers congregate to trade products, services, or assets is referred to as an economic market. While the future is unpredictable and unknowable, it is still possible to make informed predictions about the course of events. Predicting stock market movements using artificial intelligence and machine learning is one such potential. Even if the stock market is volatile, it is still feasible and wise to use artificial intelligence to create well-informed forecasts before making an investment. The current work suggests a novel approach to increase stock price forecast accuracy by integrating the Radical basis function with Particle Swarm Optimization, Slime Mold Algorithm, and Moth Flame Optimization. The objective of the study is to improve stock price forecast accuracy while accounting for the complexity and volatility of financial markets. The efficacy of the proposed strategy has been tested in the real world using historical stock price statistics. Results demonstrate considerable accuracy improvements over traditional RBF models. The combined strength of RBF and the optimization technique enhances the model's ability to adapt to changing market conditions in addition to increasing prediction accuracy. Results were 0.984, 0.990, 0.991, and 0.994 for RBF, PSO-RBF, SMA, and MFO-RBF, respectively. The performance of MFO-RBF in comparison to RBF shows how combining with the optimizer can enhance the performance of the given model. By contrasting the outcomes of various optimizers, the most accurate optimization has been determined as the main optimizer of the model.

Author 1: Zhiqiong Zou

Author 2: Guangyu Xiao

Keywords: Stock market prediction movement; prediction models; Radical basis function; optimization approaches

Paper 28: Design and Implementation of a Real-Time Image Processing System Based on Sobel Edge Detection using Model-based Design Methods

Abstract: Image processing and computer vision applications often use the Sobel edge detection technique in order to discover corners in input photographs. This is done in order to improve accuracy and efficiency. For the great majority of today's image processing applications, real-time implementation of image processing techniques like Sobel edge detection in hardware devices like field-programmable gate arrays (FPGAs) is required. Sobel edge detection is only one example. The use of FPGAs makes it feasible to have a quicker algorithmic throughput, which is required in order to match real-time speeds or in circumstances when it is critical to have faster data rates. The results of this study allowed for the Sobel edge detection approach to be applied in a manner that was not only speedy but also space-efficient. For the purpose of actually putting the recommended implementation into action, a one-of-a-kind high-level synthesis (HLS) design approach for intermediate data nodes that is based on application-specific bit widths was used. The high-level simulation code known as register transfer level (RTL) was generated by using the MATLAB HDL coder for HLS. The code written in hardware description language (HDL) that was produced was implemented on a Xilinx ZedBoard with the aid of the Vivado software, and it was tested in real time with the assistance of an input video stream.

Author 1: Taoufik Saidani

Author 2: Refka Ghodhbani

Author 3: Mohamed Ben Ammar

Author 4: Marouan Kouki

Author 5: Mohammad H Algarni

Author 6: Yahia Said

Author 7: Amani Kachoukh

Author 8: Amjad A. Alsuwaylimi

Author 9: Albia Maqbool

Author 10: Eman H. Abd-Elkawy

Keywords: Image processing; sobel edge detection; high level synthesis; model based design; Zynq7000 MATLAB HDL coder

Paper 29: Framework for Organization of Medical Processes in Medical Institutions Based on Big Data Technologies

Abstract: This research paper delves into the burgeoning field of Big Data analytics in healthcare, proposing an innovative framework aimed at refining the organization and management of medical processes within healthcare institutions. Through the lens of detailed case studies, including stroke diagnosis leveraging the UNet model, and the identification of heart and respiratory diseases via machine learning algorithms applied to data from wearable devices, the study illuminates the profound capabilities of Big Data technologies in enhancing the precision of diagnostics, tailoring patient treatment, and elevating the overall efficiency of healthcare services. It meticulously interprets the outcomes of these applications, discusses the practical implications for healthcare professionals and institutions, confronts the challenges inherent in the integration of sophisticated analytics in clinical settings, and outlines potential directions for future research. Among the pivotal challenges highlighted are issues related to data privacy, security, the need for advanced infrastructure, and the imperative for ongoing training and interdisciplinary cooperation to navigate the complexities of Big Data in healthcare. The paper underscores the transformative promise of Big Data analytics, suggesting that comprehensive adoption and adept implementation could revolutionize healthcare delivery, making it more personalized, efficient, and cost-effective. Through this exploration, the paper contributes to the ongoing discourse on the integration of technology in healthcare, offering insights into how Big Data analytics can serve as a cornerstone for the next generation of medical diagnostics and patient care management, thereby enhancing health outcomes on a global scale.

Author 1: Botagoz Zhussipbek

Author 2: Tursinbay Turymbetov

Author 3: Nuraim Ibragimova

Author 4: Zinegul Yergalauova

Author 5: Gulmira Nigmetova

Author 6: Saule Tanybergenova

Author 7: Zhanar Musagulova

Keywords: Big data; data-driven technology; artificial intelligence; medical processes; medical institutions

Paper 30: Research on Personalized Recommendation Algorithms Based on User Profile

Abstract: In recent decades, recommendation systems (RS) have played a pivotal role in societal life, closely intertwined with people's everyday activities. However, traditional recommendation systems still require thorough consideration of comprehensive user profiles as they have struggled to provide more personalized and accurate recommendation services. This paper delves into the analysis and enrichment of user profiles, utilizing this foundation to tailor recommendations for individuals across domains such as movies, TV shows, and books. The paper constructs a chart comprising 246 types of user profile attributes, primarily covering dimensions like gender, age, occupation, and religious beliefs, among 16 other dimensions. This chart integrates approximately 1.2 million data points, encompassing information relevant to movies, TV shows, and novels. Through training on the dataset, the study has enhanced the model's recommendation effectiveness. Post-training, the recommendation accuracy surpasses that of pre-training based on proposed evaluation metrics. Furthermore, post-manual evaluation, the recommended results are more reasonable and align better with user profiles.

Author 1: Guo Hui

Author 2: Zhou LiQing

Author 3: Chen Mang

Author 4: Xv ShiKun

Keywords: Recommender system; large language model; user profile; multi-disciplinary

Paper 31: Intelligent Fuzzy-PID Temperature Control System for Ensuring Comfortable Microclimate in an Intelligent Building

Abstract: In an era characterized by the growing significance of energy-efficient and human-centric environmental control systems, this research endeavors to investigate the efficacy of a Fuzzy Proportional-Integral-Derivative (PID) control approach for temperature regulation within Heating, Ventilation, and Air Conditioning (HVAC) systems. The study leverages the adaptability and robustness of fuzzy logic to dynamically tune the PID controller's parameters in response to changing environmental conditions. Through comprehensive simulations and comparative analyses, the research showcases the superior performance of the proposed fuzzy PID control system in terms of rapid response, overload avoidance, and minimal steady-state error, particularly when contrasted with conventional PID control and model predictive control (MPC) methodologies. Furthermore, the research extends its scope to assess the control system's resilience in the face of significant load variations, affirming its practical applicability in real-world HVAC scenarios. Beyond its immediate implications for HVAC systems, this research underscores the broader potential of fuzzy PID control in enhancing control precision and adaptability across various domains, including robotics, industrial automation, and process control. By advocating for future research endeavors in optimizing fuzzy membership functions, implementing real-time solutions, and exploring multi-objective optimization, among other avenues, this study seeks to contribute to the ongoing discourse surrounding advanced control strategies for achieving energy-efficient and human-centric environmental regulation.

Author 1: Rustam Abdrakhmanov

Author 2: Kamalbek Berkimbayev

Author 3: Angisin Seitmuratov

Author 4: Almira Ibashova

Author 5: Akbayan Aliyeva

Author 6: Gulira Nurmukhanbetova

Keywords: Fuzzy logic; PID; Temperature; Microclimate; Smart Building

Paper 32: Machine Learning Enhanced Framework for Big Data Modeling with Application in Industry 4.0

Abstract: In the dynamic milieu of Industry 4.0, characterized by the deluge of big data, this research promulgates a groundbreaking framework that harnesses machine learning (ML) to optimize big data modeling processes, addressing the intricate requirements and challenges of contemporary industrial domains. Traditional data processing mechanisms falter in the face of the sheer volume, velocity, and variety of big data, necessitating more robust, intelligent solutions. This paper delineates the development and application of an innovative ML-augmented framework, engineered to interpret and model complex, multifaceted data structures more efficiently and accurately than has been feasible with conventional methodologies. Central to our approach is the integration of advanced ML strategies—including but not limited to deep learning and neural networks—with sophisticated analytics tools, collectively capable of automated decision-making, predictive analysis, and trend identification in real-time scenarios. Beyond theoretical formulation, our research rigorously evaluates the framework through empirical analysis and industrial case studies, demonstrating tangible enhancements in data utility, predictive accuracy, operational efficiency, and scalability within various Industry 4.0 contexts. The results signify a marked improvement over existing models, particularly in handling high-dimensional data and facilitating actionable insights, thereby empowering industrial entities to navigate the complexities of digital transformation. This exploration underscores the potential of machine learning as a pivotal ally in evolving data strategies, setting a new precedent for data-driven decision-making paradigms in the era of Industry 4.0.

Author 1: Gulnur Kazbekova

Author 2: Zhuldyz Ismagulova

Author 3: Botagoz Zhussipbek

Author 4: Yntymak Abdrazakh

Author 5: Gulzipa Iskendirova

Author 6: Nurgul Toilybayeva

Keywords: Industry 4.0; machine learning; big data; application; management

Paper 33: Deep CNN Approach with Visual Features for Real-Time Pavement Crack Detection

Abstract: This research delves into an innovative approach to an age-old urban maintenance challenge: the timely and accurate detection of pavement cracks, a key issue linked to public safety and fiscal efficiency. Harnessing the power of Deep Convolutional Neural Networks (DCNNs), the study introduces a cutting-edge model, meticulously optimized for the nuanced task of identifying fissures in diverse pavement types, under various lighting and environmental conditions. Traditional methodologies often stumble in this regard, plagued by issues of low accuracy and high false-positive rates, predominantly due to their inability to adeptly handle the intricate variations in images caused by shadows, traffic, or debris. This paper propounds a robust algorithm that trains the model using a rich library of images, capturing an array of crack types, from hairline fractures to gaping crevices, thus imbuing the system with an astute 'understanding' of target anomalies. One salient breakthrough detailed is the model's capacity for 'context-aware' analysis, allowing for a more adaptive, precision-driven scrutiny that significantly mitigates the issue of over-generalization common in less sophisticated systems. Furthermore, the research breaks ground by integrating a novel feedback mechanism, enabling the DCNN to learn dynamically from misclassifications in an iterative refinement process, markedly enhancing detection reliability over time. The findings underscore not only improved accuracy but also heightened processing speeds, promising substantial implications for scalable real-world application and establishing a significant leap forward in predictive urban infrastructure maintenance.

Author 1: Bakhytzhan Kulambayev

Author 2: Gulnar Astaubayeva

Author 3: Gulnara Tleuberdiyeva

Author 4: Janna Alimkulova

Author 5: Gulzhan Nussupbekova

Author 6: Olga Kisseleva

Keywords: Road damage; crack; image processing; classification; segmentation

Paper 34: Development of Deep Learning Enabled Augmented Reality Framework for Monitoring the Physical Quality Training of Future Trainers-Teachers

Abstract: The fusion of augmented reality (AR) and deep learning technologies has ushered in a transformative era in the realm of real-time physical activity monitoring. This research paper introduces a system that harnesses the capabilities of PoseNet-based skeletal keypoint extraction and deep neural networks to achieve unparalleled accuracy and real-time functionality in the identification and classification of a wide spectrum of physical activities. With an impressive accuracy rate of 98% within 100 training epochs, the system proves its mettle in precise activity recognition, making it invaluable in domains such as fitness training, physical education, sports coaching, and home-based fitness. The system's real-time feedback mechanism, bolstered by AR technology, not only enhances user engagement but also motivates users to optimize their exercise routines. This paper not only elucidates the system's architecture and functionality but also highlights its potential applications across diverse fields. Furthermore, it delineates the trajectory of future research avenues, including the development of advanced feedback mechanisms, exploration of multi-modal sensing techniques, personalization for users, assessment of long-term impacts, and endeavors to ensure accessibility, inclusivity, and data privacy. In essence, this research sets the stage for the evolution of real-time physical activity monitoring, offering a compelling framework to improve fitness, physical education, and athletic training while promoting healthier lifestyles and the overall well-being of individuals worldwide.

Author 1: Sarsenkul Tileubay

Author 2: Meruert Yerekeshova

Author 3: Altynzer Baiganova

Author 4: Dariqa Janyssova

Author 5: Nurlan Omarov

Author 6: Bakhytzhan Omarov

Author 7: Zakhira Baiekeyeva

Keywords: PoseNET; MoveNET; deep learning; exercise; computer vision

Paper 35: A Cyclic Framework for Ethical Implications of Artificial Intelligence in Autonomous Vehicles

Abstract: The emergence of artificial intelligence (AI)-powered autonomous vehicles (AVs) represents a significant turning point in field of transportation, offering the potential for improved safety, efficiency, and convenience. However, the use of AI in this particular context exhibits significant ethical implications that require careful examination. This paper presents an extensive analysis of ethical considerations related integration of AI in AVs. It employs a multi-faceted approach to investigate ethical concerns of decision-making powered by AI including well-known trolley problem and moral judgments generated by AI algorithms. Additionally, it explores the complexities within safety and liability issues in the occurrence of incidents involving AVs, addressing the legal and ethical obligations of manufacturers, regulators, and users. The paper addresses the complex interaction between AI-driven transportation and its potential effects on employment and society. It provides an analysis on displacement of jobs and associated disruptions in workforce, as well as consequences for urban planning and public transportation systems. Furthermore, this study investigates the domain of privacy and data security in AVs, delving into issues related to gathering and utilization of data, as well ethical handling of personal information. Finally, this paper proposes a cyclic framework for ethical governance in AVs integrated with AI. It outlines future directions that prioritize transparency, accountability, and adherence to international humanitarian regulations. The study's findings and recommendations represent significant importance for policymakers, industry participants, and society. These stakeholders play crucial role in guiding the progress of AI in AVs, to create a transportation environment that is both safer and more ethically aligned.

Author 1: Ahmed M. Shamsan Saleh

Keywords: Artificial intelligence; autonomous vehicles; ethical implications; AI decision-making

Paper 36: Deep Learning to Predict Start-Up Business Success

Abstract: Over the past few decades, there has been rapid growth in the formation of new start-ups around the world. Thus, it is an important and challenging task to understand what makes start-ups successful and to predict their success. Several reasons are responsible for the success and failure of a start-up, including bad management, lack of funds, etc. This work aims to create a predictive model for start-ups based on many key factors involved in the early stages of a start-up’s life. Current research on predicting success mainly focuses on financial data such as ROI, revenue, etc. Therefore, in this paper, a different approach is proposed by first investigating other non-financial factors affecting start-up success and failure. Second, the adoption of an algorithm that has not been used much in predicting start-up success, which is Convolutional Neural Network (CNN). The dataset was acquired from Kaggle. The final model was reached through a series of four experiments to determine which model predicts better. The final model was implemented using a CNN with an average accuracy of 82%, an average loss of 0.4, an average 0.9 recall and an average 0.9 precision.

Author 1: Lobna Hsairi

Keywords: Deep learning; Convolutional Neural Network (CNN); prediction; start-up business

Paper 37: Data-Driven Rice Yield Predictions and Prescriptive Analytics for Sustainable Agriculture in Malaysia

Abstract: Maximizing rice yield is critical for ensuring food security and sustainable agriculture in Malaysia. This research investigates the impact of environmental conditions and management methods on crop yields, focusing on accurate predictions to inform decision-making by farmers. Utilizing machine learning algorithms as decision-support tools, the study analyses commonly used models—Linear Regression, Support Vector Machines, Random Forest, and Artificial Neural Networks—alongside key environmental factors such as temperature, rainfall, and historical yield data. A comprehensive dataset for rice yield prediction in Malaysia was constructed, encompassing yield data from 2014 to 2018. To elucidate the influence of climatic factors, long-term rainfall records spanning 1981 to 2018 were incorporated into the analysis. This extensive dataset facilitates the exploration of recent agricultural trends in Malaysia and their relationship to rice yield. The study specifically evaluates the performance of Random Forest, Support Vector Machine (SVM), and Neural Network (NN) models using metrics like Correlation Coefficient, Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Mean Squared Error (MSE), and Mean Absolute Percentage Error (MAPE). Results reveal Random Forest as the standout performer with a Correlation Coefficient of 0.954, indicating a robust positive linear relationship between predictions and actual yield data. SVM and NN also exhibit respectable Correlation Coefficients of 0.767 and 0.791, respectively, making them effective tools for rice yield prediction in Malaysia. By integrating diverse environmental and management factors, the proposed methodology enhances prediction accuracy, enabling farmers to optimize practices for better economic outcomes. This approach holds significant potential for contributing to sustainable agriculture, improved food security, and enhanced economic efficiency in Malaysia's rice farming sector. Leveraging machine learning, the research aims to transform rice yield prediction into a proactive decision-making tool, fostering a resilient and productive agrarian ecosystem in Malaysia.

Author 1: Muhammad Marong

Author 2: Nor Azura Husin

Author 3: Maslina Zolkepli

Author 4: Lilly Suriani Affendey

Keywords: Rice yield prediction; sustainable agriculture; linear regression; support vector machine; artificial neural network; predictive analytics

Paper 38: AI-based KNN Approaches for Predicting Cooling Loads in Residential Buildings

Abstract: Cooling Load (CL) estimation in residential buildings is crucial for optimizing energy consumption and ensuring indoor comfort. This article presents an innovative approach that leverages Artificial Intelligence (AI) techniques, particularly K-Nearest Neighbors (KNN), in combination with advanced optimizers, including Dynamic Arithmetic Optimization (DAO) and Wild Geese Algorithm (WGA), to enhance the accuracy of CL predictions. The proposed method harnesses the power of KNN, a machine-learning algorithm renowned for its simplicity and efficiency in regression tasks. By training on historical CL data and relevant building parameters, the KNN model can make precise predictions, 768 sample with considering factors such as Glazing Area, Glazing Area Distribution, Surface Area, Orientation, Overall Height, Wall Area, Roof Area, and Relative Compactness. Two state-of-the-art optimizers, DAO and WGA, are introduced to refine the CL estimation process further. The integration of KNN with DAO and WGA yields a robust AI-driven framework proficient in the precise estimation of CL in residential constructions. This approach not only enhances energy efficiency by optimizing cooling system operations but also contributes to sustainable building design and reduced environmental impact. Through extensive experimentation and validation, this study demonstrates the effectiveness of the proposed method, showcasing its potential to revolutionize CL estimation in residential buildings. The results indicate that the hybridization of KNN with DAO optimizers yields promising outcomes in predicting CL. The high R2 value of 0.996 and low RMSE value of 0.698 demonstrate the accuracy of the KNDA model.

Author 1: Zhaofang Du

Keywords: Cooling load; K-nearest neighbor; dynamic arithmetic optimization; wild geese algorithm

Paper 39: Enhanced Detection of COVID-19 using Deep Learning and Multi-Agent Framework: The DLRPET Approach

Abstract: The ongoing global pandemic caused by novel coronavirus (COVID-19) has emphasized the urgent need for accurate and efficient methods of detection. Over the past few years, several methods were proposed by various researchers for detecting COVID-19, but there is still a scope of improvement. Considering this, an effective and highly accurate detection model is presented in this paper that is based on Deep learning and multi-Agent concepts. Our main objective is to develop a model that can not only detect COVID-19 with high accuracy but also reduces complexity and dimensionality issues. To accomplish this objective, we applied a Deep Layer Relevance Propagation and Extra Tree (DLRPET) technique for selecting only crucial and informative features from the processed dataset. Also, a lightweight ResNet based Deep Learning model is proposed for classifying the disease. The ResNet model is initialized three times creating agents which analyses the data individually. The novelty contribution of this work is that instead of passing the entire training set to the classifier, we have divided the training dataset into three subsets. Each subset is passed to a specific agent for training and making individual predictions. The final prediction in proposed network is made by implementing majority voting mechanism to determine whether an individual is COVID-19 positive or negative. The experimental outcomes indicated that our approach achieved an accuracy of 99.73% that is around 2% higher than standard best performing KISM model. Moreover, proposed model attained precision of 100%, recall of 99.73% and F1-score of 98.59 % respectively, showing an increase of 5% in precision, 4.73% in recall and 4.59% in F1-score than best performing SVM model.

Author 1: Rupinder Kaur Walia

Author 2: Harjot Kaur

Keywords: COVID-19; deep learning; SVM; ResNet; disease classification; biomedical applications; multi-agent

Paper 40: Experimental IoT System to Maintain Water Quality in Catfish Pond

Abstract: This study investigates the challenges in catfish aquaculture, mainly focusing on water quality, which is crucial for successful fish farming. This research aims to implement Internet of Things (IoT) technology with sensors connected to a microcontroller to monitor and control critical parameters such as temperature, pH, and oxygen levels in catfish ponds. Utilizing NodeMCU and specific sensors, the system provides real-time monitoring, enabling early detection of environmental changes that could impact fish health. The research findings indicate IoT technology in catfish aquaculture can enhance fish health and growth. Real-time monitoring reduces the risk of diseases by providing an optimal environment for the fish. Additionally, automatic control using fuzzy logic, which can adjust email notifications automatically, and actuators such as water pumps and pH regulators that work automatically based on conditions help maintain the stability of water quality. A comparison between conventional and IoT-based farming reveals that the IoT system can reduce catfish mortality by optimizing feed distribution and regulating pH levels. Thus, this study positively contributes to developing more efficient, sustainable and healthy catfish aquaculture methods through implementing IoT technology.

Author 1: Adani Bimasakti Wibisono

Author 2: Riyanto Jayadi

Keywords: IoT; aquaculture; catfish cultivation; monitoring; controlling

Paper 41: Revolutionizing Education: Cutting-Edge Predictive Models for Student Success

Abstract: Student performance prediction systems are crucial for improving educational outcomes in various institutions, including universities, schools, and training centers. These systems gather data from diverse sources such as examination centers, registration departments, virtual courses, and e-learning platforms. Analyzing educational data is challenging due to its vast and varied nature, and to address this, machine learning techniques are employed. Dimensionality reduction, enabled by machine learning algorithms, simplifies complex datasets, making them more manageable for analysis. In this study, the Support Vector Classification (SVC) model is used for student performance prediction. SVC is a powerful machine-learning approach for classification tasks. To further enhance the model's efficiency and accuracy, two optimization algorithms, the Sea Horse Optimization (SHO) and the Adaptive Opposition Slime Mould Algorithm (AOSMA), are integrated. Machine learning (ML) reduces complexity through techniques like feature selection and dimensionality reduction, improving the effectiveness of student performance prediction systems and enabling data-informed decisions for educators and institutions. The combination of SVC with these innovative optimization strategies highlights the study's commitment to leveraging the latest advancements in ML and bio-inspired algorithms for more precise and robust student performance predictions, ultimately enhancing educational outcomes. Based on the obtained outcomes, it reveals that the SVSH model registered the best performance in predicting and categorizing the student performance with Accuracy=92.4%, Precision=93%, Recall=92%, and F1_Score=92%. Implementing SHO and AOSMA optimizers to the SVC model resulted in improvement of Accuracy evaluator outputs by 2.12% and 0.89%, respectively.

Author 1: Moyan Li

Author 2: Suyawen

Keywords: Student performance; Support Vector Classification; sea horse optimization; adaptive opposition slime mould algorithm

Paper 42: Disease-Aware Chest X-Ray Style GAN Image Generation and CatBoost Gradient Boosted Trees

Abstract: Artificial Intelligence has significantly advanced and is proficient in image classification. Even though the COVID-19 pandemic has ended, the virus is now considered to have entered an endemic phase. Historically, COVID-19 detection has predominantly depended on a single technology known as the polymerase chain reaction (PCR). The academic community is keen radiograph data to forecast COVID-19 because of its prospective advantages. The proposed methodology aims to improve dataset quality by utilizing artificially generated images produced by StyleGAN. The ratio of 59:41 was used to combine the synthetic datasets with the real ones. The combination of the StyleGAN framework, the VGG19, and CatBoost Gradient Boosted Trees is to improve prediction accuracy. Accurate and precise measurements significantly impact the evaluation of a model's performance. The assessment resulted in 98.67% accurate and 97.21% precise. In the future, we may enhance the diversity and quality of the collection by integrating other datasets from different sources with the Chest X-ray dataset.

Author 1: Andi Besse Firdausiah Mansur

Keywords: Artificial intelligence; StyleGAN; chest X-ray prediction; COVID19; CatBoost gradient boosted trees

Paper 43: Cyber Security Intrusion Detection and Bot Data Collection using Deep Learning in the IoT

Abstract: In the digital age, cybersecurity is a growing concern, especially as IoT continues to grow rapidly. Cybersecurity intrusion detection systems are critical in protecting IoT environments from malicious activity. Deep learning approaches have emerged as promising intrusion detection techniques due to their ability to automatically learn complex patterns and features from large-scale data sets. In this research, we give a detailed assessment of the use of deep learning algorithms for cybersecurity intrusion detection in IoT contexts. The study discusses the challenges of securing IoT systems, such as device heterogeneity, limited computational resources, and the dynamic nature of IoT networks. To detect intrusions in IoT environments, convolutional neural networks (CNNs) and recurrent neural networks (RNNs) have been used. The NF-UQ-NIDS and NF-Bot-IoT data sets are used for training and assessing deep learning-based intrusion detection systems. Our study also explores using deep learning approaches to identify botnets in IoT settings to counter the growing threat of botnets. Also, analyze representative bot data sets and explain their significance in understanding botnet behavior and effective defenses. The study evaluated IDS performance and traffic flow in the IoT context using various machine learning algorithms. For IoT environments, the results highlight the importance of selecting appropriate algorithms and employing effective data pre-processing techniques to improve accuracy and performance. Cyber-attack detection with the proposed system is highly accurate when compared with other algorithms for both NF-UQ-NIDS and NF-BoT-IoT data sets.

Author 1: Fahad Ali Alotaibi

Author 2: Shailendra Mishra

Keywords: Internet of things; intrusion detection system; random neural networks; feed forward neural networks; convolutional neural networks

Paper 44: Method for Disaster Area Detection with Just One SAR Data Acquired on the Day After Earthquake Based on YOLOv8

Abstract: Method for earthquake disaster area detection with just a single satellite-based SAR data which is acquired on the day after earthquake based on object detection method of YOLOv8 and Detectron2 is proposed. Through experiments with several SAR data derived from the different SAR satellites which observed Noto Peninsula earthquake occurred on the first of January 2024, it is found that the proposed method works well to detect several types of damages effectively. Also, it is found that the proposed method based on “Roboflow” and YOLOv8 as well as Detectron2 for annotation and object detection is appropriate for disaster area detection. Furthermore, it is possible to detect disaster areas even if just one single SAR data which acquired on the day after the disaster occurred because the trained learning model for disaster area detection is created through experiments.

Author 1: Kohei Arai

Author 2: Yushin Nakaoka

Author 3: Hiroshi Okumura

Keywords: SAR; YOLOv8; Detectron2; earthquake; disaster; disaster area detection; noto peninsula earthquake

Paper 45: Forecasting the Yoga Influence on Chronic Venous Insufficiency: Employing Machine Learning Methods

Abstract: This investigation introduces a groundbreaking approach to unravel the complexities of Chronic Venous Insufficiency (CVI) by leveraging machine learning, notably the Support Vector Classification (SVC), alongside optimization systems like Dwarf Mon-goose Optimization (DMO) and Smell Agent Optimization (SAO). This pioneering strategy not only aims to bolster predictive Precision but also seeks to optimize personalized treatment paradigms for CVI, presenting a compelling avenue for the advancement of healthcare solutions. The study aims to predict the impact of yoga on CVI using a comprehensive dataset, incorporating demographic information, baseline severity indicators, and yoga practice details. Through meticulous feature engineering, machine learning algorithms forecast outcomes such as changes in symptom severity and overall well-being improvements. This predictive model has the potential to transform personalized CVI treatment plans by offering tailored recommendations for specific yoga practices, optimizing therapeutic approaches, and guiding efficient healthcare resource allocation. Ethical considerations, patient preferences, and safety are highlighted for responsible translation into clinical settings. The integration of SVC with optimization systems presents a novel and promising approach, contributing meaningfully to personalized CVI management and providing valuable insights for current and future practices. The results obtained for VCSS-PRE and VCSS-1 unequivocally highlight the outstanding performance of the SVDM model in both prediction and categorization. The model achieved remarkable Accuracy and Precision values, attaining 92.9% and 93.1% for VCSS-PRE and 94.3% and 94.9% for VCSS-1.

Author 1: Xiao Du

Keywords: Chronic Venous Insufficiency; yoga; classification; machine learning; Support Vector Classification; smell agent optimization; Dwarf Mongoose Optimization

Paper 46: Multi-Track Music Generation Based on the AC Algorithm and Global Value Return Network

Abstract: In the current field of deep learning and music information retrieval, automated music generation has become a hot research topic. This study addresses the issues of low clarity and musicality in current multi-track music generation by combining the Actor-Critic algorithm and the Global Value Return Network to create a novel multi-track music generation model. The study first utilizes the Actor-Critic algorithm to generate single-track music rhythm and melody models. Building upon this foundation, the study further optimizes the single-track models using the Global Value Return Network and proposes the multi-track music model. The results demonstrate that the harmonization accuracy of the final multi-track music generation model ranges from 0.90 to 0.98, with a maximum value of 0.98. Additionally, the audience satisfaction and expert satisfaction of the model are 0.96 and 0.97, respectively, indicating that the model has a high musical appreciation value. Overall, the multi-track music generation model designed in this study addresses the limitations of single-track music generation and produces more rhythmically diverse multi-track music.

Author 1: Wei Guo

Keywords: AC; global value; return network; track; music model; rhythm; melody

Paper 47: Defining Integrated Agriculture Information System Non-Functional Requirement and Re-engineering the Metadata

Abstract: Developing a well-functioning information system like integrated agriculture information system (IAIS) requires a list of task requirements that will be transformed into system features. Feature Driven Development (FDD) model is suitable for this situation. The requirements for building an information system are not solely based on functional needs but also non-functional requirements (NFR). Non-functional requirements also play a crucial role in system development as they affect business process management. A well-defined business process will ultimately result in robust system features. It is essential to map non-functional requirements to the business process to clearly identify the information system requirements that will become new features. Not only can NFR enrich system metadata and databases, but they also serve as the initial foundation for the system coding process, leading to the final information system output. This study creates a flow diagram mapping NFR to the business process using Business Process Management Notation (BPMN). Several identified NFR categories are then transformed into metadata and use case diagrams. The formation of this NFR mapping flow diagram is expected to facilitate information system development by visualizing system requirements in a forward and backward flow according to the sequence of processes. Feature development can be streamlined in the event of NFR changes by tracing NFR and related features.

Author 1: Argo Wibowo

Author 2: Antonius Rachmat Chrismanto

Author 3: Gabriel Indra Widi Tamtama

Author 4: Rosa Delima

Keywords: Information system; non-functional requirements; BPMN; metadata; feature-driven development

Paper 48: Novel Design of a Robotic Arm Prototype with Complex Movements Based on Surface EMG Signals to Assist Disabilities in Vietnam

Abstract: In recent years, surface electromyography (sEMG) signals have been recognized as a type of signal with significant practical implications not only in medicine but also in the field of science and engineering for functional rehabilitation. This study focuses on understanding the application of surface electromyography signals in controlling a robotic arm for assisting disabled individuals in Vietnam. The raw sEMG signals, collected using appropriate sensors, have been processed using an effective method that includes several steps such as A/D converting and the use of band-pass and low-pass filters combined with an envelope detector. To demonstrate the meaningful effectiveness of the processed sEMG signals, the study has designed a robotic arm model with complex finger movements similar to those of a human. The experimental results show that the robotic arm operates effectively, with fast response times, meeting the support needs of disabled individuals.

Author 1: Ngoc–Khoat Nguyen

Author 2: Thi–Mai–Phuong Dao

Author 3: Van–Kien Nguyen

Author 4: Van–Hung Pham

Author 5: Van–Minh Pham

Author 6: Van–Nam Pham

Keywords: Disabilities; sEMG; signal processing; human arm; robotic arm

Paper 49: A Bloom Cognitive Hierarchical Classification Model for Chinese Exercises Based on Improved Chinese-RoBERTa-wwm and BiLSTM

Abstract: Assessing students' cognitive ability is one of the most important prerequisites for improving learning effectiveness, and the process involves aspects such as exercises, students' answers and teaching cases. In order to effectively assess students' cognitive ability, this paper proposes a Chinese text classification model that can automatically and accurately classify Bloom's cognitive hierarchy of exercises, starting from the exercises. Firstly, FreeLB perturbation is added to the input Embedding to enhance the generalization performance of the model, and Chinese-RoBERTa-wwm is used to obtain the pooler information and sequence information of the text; secondly, LSTM is used to extract the deep-associative features in the sequence information and combine with the pooler information to construct the semantically informative word vectors; lastly, the word vectors are fed into BiLSTM to learn the sequence bi-directional dependency information to obtain more comprehensive semantic features to achieve the accurate classification of the exercises. Experiments show that the model proposed in this paper significantly outperforms the baseline model on three Chinese public datasets, achieving 94.8%, 94.09% and 94.71% accuracies respectively, and also effectively performs the Bloom cognitive hierarchy classification task on two Chinese exercise datasets with less data.

Author 1: Zhaoyu Shou

Author 2: Yipeng Liu

Author 3: Dongxu Li

Author 4: Jianwen Mo

Author 5: Huibing Zhang

Keywords: Chinese Text Classification; Chinese-RoBERTa-wwm; BiLSTM; Bloom Cognitive Hierarchy

Paper 50: The Application of Improved Scale Invariant Feature Transformation Algorithm in Facial Recognition

Abstract: Currently, face recognition models suffer from insufficient accuracy, stability, and computational efficiency. To address this issue, an improved feature extraction algorithm on the ground of Haar wavelet features and scale invariant feature transformation algorithm is proposed. In addition, the study also combines this algorithm with deep belief networks to construct an improved facial recognition model. The effectiveness of the proposed improved feature extraction algorithm was verified, and it was found that the recognition accuracy of the algorithm was 94.2%, which is better than other comparative algorithms. In addition, the study also conducted empirical analysis on the improved facial recognition model and found that the recognition accuracy of the model was 0.92, and the feature matching time was 2.6 seconds, which was better than other comparative models in terms of performance. On the ground of the above results, the proposed facial recognition model has significantly improved recognition accuracy and efficiency compared to traditional models. It can provide theoretical reference for improving the universality of facial recognition applications in different fields.

Author 1: Yingzi Cong

Keywords: Haar wavelet features; scale invariant feature transformation algorithm; deep belief network; facial recognition; performance improvement

Paper 51: Designing a Mobile Application for Identifying Strawberry Diseases with YOLOv8 Model Integration

Abstract: The progress in computer vision has led to the development of potential solutions, becoming a versatile technological key to addressing challenging issues in agriculture. These solutions aim to enhance the quality of agricultural products, boost the economy's competitiveness, and reduce labor and costs. Specifically, the detection of diseases in various fruits before harvest to avoid reducing product quality and quantity still relies on the experience of long-time farmers. This leads to difficulties in controlling disease sources over large cultivated areas, resulting in uneven quality control after harvest, which may lead to low prices or failure to meet export requirements to developed markets. Therefore, this stage has now been applied with modern technology to gradually replace humans. In this paper, we propose a mobile application to detect four common diseases in strawberry trees by using image processing technology that combines an artificial intelligence network in identification: based on size, color, and shape defects on the surface of the fruit. The proposed model consists of different versions of YOLOv8 with RGB input to accurately detect diseases in strawberries and provide assessments. Among these, the YOLOv8n model utilizes the fewest parameters with only 11M, but it produces more output parameters with higher accuracy compared to some other YOLOv8 models, achieving an average accuracy of approximately 87.9%. Therefore, the proposed method emerges as one of the possible solutions for strawberry disease detection.

Author 1: Thuy Van Tran

Author 2: Quang - Huy Do Ba

Author 3: Kim Thanh Tran

Author 4: Dang Hai Nguyen

Author 5: Dinh Chung Dang

Author 6: Van - Luc Dinh

Keywords: Computer vision; YOLOv8; strawberry diseases

Paper 52: Exploring the Landscape: Analysis of Model Results on Various Convolutional Neural Network Architectures for iRESPOND System

Abstract: In the era of rapid technological advancement, the integration of cutting-edge technologies plays a pivotal role in enhancing the efficiency and responsiveness of critical systems. iRESPOND, a real-time Geospatial Information and Alert System, stands at the forefront of such innovations, facilitating timely and informed decision-making in dynamic environments. As the demand for accurate and swift responses, the role of CNN models in iRESPOND becomes significant. The study focuses on seven prominent CNN architectures, namely EfficientNet (B0, B7, V2B0, and V2L), InceptionV3, ResNet50, and VGG19 and with the integration of different optimizers and learning rates. The methodology employed a strategic implementation of looping during the training phase. This iterative approach is designed to systematically re-train the CNN models, emphasizing identifying the most suitable architecture among the seven considered variants. The primary objective is to discern the optimal architecture and fine-tune critical parameters, explicitly targeting the optimizer and learning rate values. The differential impact of each model on the system's ability is to discern patterns and anomalies in the image datasets. ResNet50 exhibited robust performance showcasing suitability for real-time processing in dynamic environments with a better accuracy result of 95.02%. However, the EfficientNetV2B0 model, characterized by its advancements in network scaling, presented promising results with a lower loss of 0.187. Generally, the findings not only contribute valuable insights into the optimal selection of architectures for iRESPOND but also highlight the importance of fine-tuning hyperparameters through an iterative training approach, which paves the way for the continued enhancement of iRESPOND as an adaptive system.

Author 1: Freddie Prianes

Author 2: Kaela Marie Fortuno

Author 3: Rosel Onesa

Author 4: Brenda Benosa

Author 5: Thelma Palaoag

Author 6: Nancy Flores

Keywords: Artificial intelligence; image classification; emergency response; model training; optimizers; learning rate

Paper 53: Performance Analysis for Secret Message Sharing using Different Levels of Encoding Over QSDC

Abstract: It was recently proposed to use quantum secure direct communication (QSDC), a branch of quantum cryptography, to secure data transfers from sender to receiver without relying on computational complexity. Despite the benefits of multiphoton, sending secret messages between several parties in a quantum channel still presents a challenge because the current multiphoton only considers two parties. When more parties are included, the scalability problem becomes apparent. Therefore, the scalable multiphoton approach is needed to allow secure sharing between the legal parties. The manipulation of level encoding provides new opportunities for more efficient quantum information processing and message sharing. This research aims to propose a strategy that uses four-level encoding with the multiphoton approach to share secret messages between multi-party. From the analysis conducted, it has been shown that a high number of level encoding can shorten the time taken for photon transmission between parties and an attacker has a lower probability of chances to launch an attack, however, communication will be affected due to high sensitivity to noise.

Author 1: Nur Shahirah Binti Azahari

Author 2: Nur Ziadah Binti Harun

Author 3: Chai Wen Chuah

Author 4: Rosmamalmi Mat Nawi

Author 5: Zuriati Binti Ahmad Zukarnain

Author 6: Nor Iryani Binti Yahya

Keywords: Multiphoton approach; multi-party; level of encoding; scalability; error probability

Paper 54: Handling Transactional Data Features via Associative Rule Mining for Mobile Online Shopping Platforms

Abstract: Transactional data processing is often a reflection of a consumer's buying behavior. The relational records if properly mined, helps business managers and owners to improve their sales volume. Transaction datasets are often rippled with the inherent challenges in their manipulation, storage and handling due to their infinite length, evolution of product features, evolution in product concept, and oftentimes, a complete drift away from product feat. The previous studies' inability to resolve many of these challenges as abovementioned, alongside the assumptions that transactional datasets are presumed to be stationary when using the association rules – have been found to also often hinder their performance. As it deprives the decision support system of the needed flexibility and robust adaptiveness to manage the dynamics of concept drift that characterizes transaction data. Our study proposes an associative rule mining model using four consumer theories with RapidMiner and Hadoop Tableau analytic tools to handle and manage such large data. The dataset was retrieved from Roban Store Asaba and consists of 556,000 transactional records. The model is a 6-layered framework and yields its best result with a 0.1 value for both the confidence and support level(s) at 94% accuracy, 87% sensitivity, 32% specificity, and a 20-second convergence and processing time.

Author 1: Maureen Ifeanyi Akazue

Author 2: Sebastina Nkechi Okofu

Author 3: Arnold Adimabua Ojugo

Author 4: Patrick Ogholuwarami Ejeh

Author 5: Christopher Chukwufunaya Odiakaose

Author 6: Frances Uche Emordi

Author 7: Rita Erhovwo Ako

Author 8: Victor Ochuko Geteloma

Keywords: Association rule mining; online shopping platforms; feature evolution; concept drift; concept evolution; shelf placement

Paper 55: An Approach for Developing an Ontology: Learned from Business Model Ontology Design and Development

Abstract: Ontology, serving as an explicit specification of conceptualization, has found widespread applications across various fields. Business Model Ontology (BMO) stands out as a prominent ontology, especially in the domains of business and entrepreneurship. This study employs the narrative literature review method to delve into the Ontology Development Method (ODM). By identifying commonalities among various ODMs and drawing insights from the BMO, the study proposes a Unified Ontology Approach (UOA) as an alternative ODM. The UOA is derived by combining the common characteristics and key steps of various ODMs, aiming to streamline the ontology development process and enhance its effectiveness. Through an extensive analysis of existing methodologies, this research contributes to the field by offering a consolidated perspective on ODMs. The study findings shed light on the strengths and weaknesses of different approaches, facilitating informed decision-making for ontology developers. Furthermore, the discussion explores the implications of adopting the UOA in practical applications, emphasizing its potential to improve ontology quality, interoperability, and adaptability across diverse domains. In conclusion, this paper advocates for the adoption of the UOA as a comprehensive and flexible framework for ontology development. By synthesizing the strengths of existing ODMs and insights from the BMO, the UOA offers a promising avenue for advancing the field of ontology development and driving progress in various domains and applications.

Author 1: Ahadi Haji Mohd Nasir

Author 2: Mohd Firdaus Sulaiman

Author 3: Liew Kok Leong

Author 4: Ely Salwana

Author 5: Mohammad Nazir Ahmad

Keywords: Ontology; Ontology Development Method (ODM); Business Model Ontology (BMO); Unified Ontology Approach (UOA)

Paper 56: Deep Convolutional Neural Networks Fusion with Support Vector Machines and K-Nearest Neighbors for Precise Crop Leaf Disease Classification

Abstract: Maize and Paddy are pivotal crops in India, playing a vital role in ensuring food security. Timely detection of diseases and the implementation of remedial measures are crucial for securing optimal crop yield and profitability for farmers. This study utilizes a dataset encompassing images of diseased maize and paddy leaves, addressing various conditions such as corn blight, common rust, gray leaf spot, brown spot, hispa, and leaf blast, alongside images of healthy leaves. The dataset used here is a combination of online repository as well as manually collected samples from neighborhood farmlands at different growth stages. A machine vision approach that is accessible, quick, robust and cost effective to determine crop leaf diseases is need of the hour. In the proposed work, using transfer-learning approach, many Deep Convolutional Neural Networks (DCNN) and hybrid DCNNs have been developed, trained, validated and tested. To achieve better accuracy, integration of DCNNs and machine learning classifiers like multiclass Support Vector Machine (SVM) and K-Nearest Neighbor (KNN) algorithms is carried out. The research is carried out in four stages, in the first stage, DCNNs have been used as classifiers. Subsequently, these same DCNNs are repurposed as feature extractors, and the extracted features are input into classifiers such as multiclass SVM and KNN. In the third stage, an ensemble of DCNNs is performed for networks exhibiting excellent performance during first stage. At a fourth stage, features extracted from these ensemble networks are fed into the same multiclass SVM and KNN classifiers to assess accuracy. A total of 1600 images for training and 400 images for testing are used. For maize data set, we achieved a 100% accuracy in AlexNet plus VGG-16 hybrid network for multiclass SVM with 75:25 split ratio and for paddy dataset 99.51% accuracy is achieved in ResNet-50 plus Darknet-53 hybrid network for multiclass SVM with 75:25 split ratio. In the proposed study a comprehensive analysis is conducted, exploring features from various layers and adjusting data split ratios.

Author 1: Sunil Kumar H R

Author 2: Poornima K M

Keywords: Deep Convolutional Neural Network (DCNN); multiclass Support Vector Machine (SVM); K-Nearest Neighbor (KNN); ensemble; features; accuracy

Paper 57: Towards a Machine Learning-based Model for Corporate Loan Default Prediction

Abstract: As the core business of the banking system is to lend money and then get it back, loan default is one of the most crucial issues for commercial banks. With data analysis and artificial intelligence, extracting valuable information from historical data, to lower their losses, banks would be able to classify their customers and predict the probability of credit repayment instead of relying on traditional methods. As most actual research is focused on individuals’ loans, the novelty of the present paper is to treat corporate loans. Its main objective is to propose a model to address the problem using selected machine learning algorithms to classify companies into two classes to be able to predict loan defaulters. This paper delves into the Corporate Loan Default Prediction Model (CLD PM), which is designed to forecast loan defaults in corporations. The model is grounded in the CRISP-DM process, commencing with comprehending corporate requirements and implementing classification techniques. The data acquisition and preparation phase are critical in testing the selected algorithms, which involve Logistic Regression, Decision Tree, Support Vector Machine, Random Forest, XGBoost, and Adaboost. The model's efficacy is assessed using various metrics, namely Accuracy, Precision, Recall, F1 score, and AUC. Subsequently, the model is scrutinized using an actual dataset of loans for Moroccan real estate firms. The findings reveal that the Random Forest and XGBoost algorithms outperformed the others, with every metric surpassing 90%. This was accomplished by utilizing SMOTE as an oversampling method, given the dataset's imbalance. Furthermore, when concentrating on financial statements, selecting the five most significant financial ratios and the company's age, Random Forest was adept at predicting defaulters with good results: accuracy of 90%, precision of 75%, recall of 50%, F1 score of 60% and AUC of 77%.

Author 1: Imane RHZIOUAL BERRADA

Author 2: Fatimazahra BARRAMOU

Author 3: Omar BACHIR ALAMI

Keywords: Loan default; prediction; artificial intelligence; data analysis; machine learning; companies; corporate; real estate; bank

Paper 58: Enhancing Data Warehouses Security

Abstract: Data Warehouses (DWs) are essential for enterprises, containing valuable business information and thus becoming prime targets for internal and external attacks. Data warehouses are crucial assets for organizations, serving critical purposes in business and decision-making. They consolidate data from diverse sources, making it easier for organizations to analyze and derive insights from their data. However, as data is moved from one source to another, security issues arise. Unfortunately, current data security solutions often fail in DW environments due to resource-intensive processes, increased query response times, and frequent false positive alarms. The structure of the data warehouse is designed to facilitate efficient analysis. Developing and deploying a data warehouse is a difficult process and its security is an even greater concern. This study provides a comprehensive review of existing data security methods, emphasizing their implementation challenges in DW environments. Our analysis highlights the limitations of these solutions, particularly in meeting scalability and performance needs. We conclude that current methods are impractical for DW systems and support for a comprehensive solution tailored to their specific requirements. Our findings underscore the ongoing significance of data warehouse security in industrial projects, necessitating further research to address remaining challenges and unanswered questions.

Author 1: Muhanad A. Alkhubouli

Author 2: Hany M. Lala

Author 3: AbdAllah A. AlHabshy

Author 4: Kamal A. ElDahshan

Keywords: Data warehouse; data security; encryption; security issues; data integrity; privacy; confidentiality

Paper 59: Speech Emotion Recognition in Multimodal Environments with Transformer: Arabic and English Audio Datasets

Abstract: Speech Emotion Recognition (SER) is a fast-developing area of study with a primary goal of automatically identifying and analyzing the emotional states expressed in speech. Emotions are crucial in human communication as they impact the effectiveness and meaning of linguistic expressions. SER aims to create computational approaches and models to detect and interpret emotions from speech signals. One of the primary applications of SER is evident in the field of Human-Computer Interaction (HCI), where it can be used to develop interactive systems that adapt to the user's emotional state based on their voice. This paper investigates the use of speech data for speech emotion recognition. Additionally, we applied a transformation process to convert the speech data into 2D images. Subsequently, we compared the outcomes of this transformation with the original speech data, aligning the comparison with a dataset containing labeled speech samples in both Arabic and English. Our experiments compare three methods: a transformer-based model, a Vision Transformer (ViT) based model, and a wave2vec-based model. The transformer model is trained from scratch on two significant audio datasets: the Arabic Natural Audio Dataset (ANAD) and the Toronto Emotional Speech Set (TESS), while the vision transformer is evaluated alongside wave2vec as part of transfer learning. The results are impressive. The transformer model achieved remarkable accuracies of 94% and 99% on ANAD and TESS datasets, respectively. Additionally, ViT demonstrates strong capabilities, achieving accuracies of 88% and 98% on the ANAD and TESS datasets, respectively. To assess the transfer learning potential, we also explore the Wave2Vector model with fine-tuning. However, the findings suggest limited success, achieving only a 56% accuracy rate on the ANAD dataset.

Author 1: Esraa A. Mohamed

Author 2: Abdelrahim Koura

Author 3: Mohammed Kayed

Keywords: Speech emotion recognition; transformer encoder; fine-tuning; wav2vec; multimodal emotion recognition

Paper 60: A Method for Constructing and Managing Level of Detail for Non-Closed Boundary Models of Buildings

Abstract: An urban digital twin (UDT) involves creating a virtual three-dimensional (3D) digital replica of a real-world city. To build a UDT model, it needs a comprehensive 3D representation of the city's terrain, buildings, and infrastructure. In order to effectively visualize and manage large-scale spatial data in 3D, it is essential to establish and maintain an appropriate level of detail (LoD) for the 3D model. This study proposes to construct and manage LoDs for VWorld building data. However, since buildings are often composed of non-closed boundary models, applying a quadric mesh-based simplification algorithm may result in the deletion of meshes containing important contour information that defines the shape of the building. To overcome this problem, this paper proposes to use a geometric filtering algorithm to preserve the building outline shape.

Author 1: Ahyun Lee

Keywords: GIS; digital twin; 3D map; level of detail

Paper 61: The Optimal Allocation Method for Energy Storage in Low Voltage Distribution Power Network

Abstract: In order to promote the absorption of photovoltaic in low-voltage distribution network, and reduce the voltage over-limit problem caused by high proportion of distributed photovoltaics, this paper proposes a method for optimizing the allocation of distributed energy storage system in low voltage distribution network. Firstly, based on the node voltage of the maximum load day and all day, the optimal clustering number k is obtained by the elbow method, and the K-means clustering algorithm is used to realize the zoning of the distribution network. Secondly, the objective function is to improve the node voltage, reduce the power loss, and minimize the comprehensive cost of energy storage investment, and at the same time consider various constraints such as power balance and energy storage battery, and construct a multi-objective optimization model for the optimal configuration of distributed energy storage system in low voltage distribution network. After normalizing each objective function, the weight coefficient of each objective function is determined based on the analytic hierarchy method. The whale algorithm is used to solve the model to determine the best installation location and capacity of distributed energy storage. Finally, taking an actual area as an example, the effectiveness of the proposed model in leveling the voltage exceeding of low voltage distribution network nodes is verified.

Author 1: Lin Zhu

Author 2: Xiaofang Meng

Author 3: Nannan Zhang

Keywords: Optimal allocation; voltage over-limit; distributed energy storage; low voltage distribution networks

Paper 62: Enhancing Harris Hawks Optimization Algorithm for Resource Allocation in Cloud Computing Environments

Abstract: Cloud computing is revolutionizing the delivery of on-demand scalable and customizable resources. With its flexible resource access and diverse service models, cloud computing is essential to modern computing infrastructure. In cloud environments, assigning Virtual Machines (VMs) to Physical Machines (PMs) remains a complex and challenging task critical to optimizing resource utilization and minimizing energy consumption. Given the NP-hard nature of VM allocation, solving this optimization problem requires efficient strategies, usually addressed by metaheuristic algorithms. This study introduces a novel method for allocating VMs based on the Harris Hawks Optimization (HHO) algorithm. HHO has exhibited the capacity to provide optimal solutions to specific issues inspired by the hunting behavior of Harris's falcons in the natural world. However, there are often problems with convergence to local optima, which affects the quality of the solution. To mitigate this challenge, this study employs a tent chaotic map during the initialization phase, aiming for enhanced diversity in the initial population. The proposed method, Enhanced HHO (EHHO), has superior performance compared to previous algorithms. The results confirm the effectiveness of the introduced tent chaotic map improvement and suggest that EHHO can improve solution quality, higher convergence speed, and improved robustness in addressing VM allocation challenges in cloud computing deployments.

Author 1: Ganghua Bai

Keywords: Cloud computing; virtual machine allocation; energy efficiency; resource utilization

Paper 63: Hybrid Machine Learning Approaches for Predicting and Diagnosing Major Depressive Disorder

Abstract: Major Depressive Disorder (MDD) is common and debilitating, requiring accurate prediction and diagnosis. This study uses clinical, demographic, and EEG data to test hybrid machine learning methods for MDD prediction and diagnosis. EEG data reveals brain electrical activity and can identify MDD patterns and traits. The study aimed to enhance Major Depressive Disorder (MDD) prediction and diagnosis using hybrid machine learning methods, focusing on EEG data alongside clinical and demographic information. Employing various algorithms like CatBoost, Random Forest, XG Boost, XGB Random Forest, SVM with a linear kernel, and logistic regression with Elasticnet regularization, the study found that CatBoost achieved the highest accuracy of 93.1% in MDD prediction and diagnosis, surpassing other models. Additionally, the ensemble model combining XGBoost and Random Forest showed strong performance in ROC analysis, effectively discriminating between individuals with and without MDD. These findings underscore the potential of EEG data integration and hybrid machine learning techniques in accurately identifying and classifying MDD patients, paving the way for personalized interventions and targeted treatments in depressive disorders.

Author 1: N. Balakrishna

Author 2: M. B. Mukesh Krishnan

Author 3: D. Ganesh

Keywords: Major Depressive Disorder (MDD); hybrid machine learning; cat boost; random forest; XG boost; XGB random forest; SVM; logistic regression; EEG data

Paper 64: Enhance Telecommunication Security Through the Integration of Support Vector Machines

Abstract: This research investigates the escalating issue of telephone-based fraud in Indonesia, a consequence of enhanced connectivity and technological advancements. As the telecommunications sector expands, it faces increased threats from sophisticated criminal activities, notably voice call fraud, which leads to significant financial losses and diminishes trust in digital systems. This study presents a novel security system that leverages the capabilities of Support Vector Machines (SVM) for the advanced classification of complex patterns inherent in fraudulent activities. By integrating SVM algorithms, this system aims to effectively process and analyze large volumes of data to identify and prevent fraudulent acts. The utilization of SVM in our proposed framework represents a significant strategy to combat the adaptive and evolving tactics of cybercriminals, thereby bolstering the resilience of telecommunications infrastructure. Upon further refinement, the system exhibited a substantial improvement in identifying fraudulent activities, with accuracy rates increasing from 81% to 86%. This enhancement underscores the system's efficacy in real-world scenarios. Our research underscores the critical need to marry technological innovations with ethical and privacy considerations, highlighting the role of public awareness and education in augmenting security measures. The development of this SVM-based security system constitutes a pivotal step towards reinforcing Indonesia's telecommunications infrastructure, contributing to the national objective of securing the digital economy and fostering a robust digital ecosystem. By addressing current and future cyber threats, this approach exemplifies Indonesia's commitment to leveraging technology for societal welfare, ensuring a secure and prosperous digital future for its citizens.

Author 1: Agus Tedyyana

Author 2: Adi Affandi Ahmad

Author 3: Mohd Rushdi Idrus

Author 4: Ahmad Hanis Mohd Shabli

Author 5: Mohamad Amir Abu Seman

Author 6: Osman Ghazali

Author 7: Jaroji

Author 8: Abd Hadi Abd Razak

Keywords: Call security system; artificial intelligence; support vector machine; data analysis; fraud detection system

Paper 65: Virtual Reality and Augmented Reality in Artistic Expression: A Comprehensive Study of Innovative Technologies

Abstract: Over the last decade, Virtual Reality (VR) and Augmented Reality (AR) have gained popularity across various industries, particularly the arts, thanks to technological advances and inexpensive hardware and software availability. These technologies have redefined the boundaries of creativity and immersive experiences in artistic expression. This paper explores the dynamic interface between AR, VR, and the diverse Information Technology (IT) landscape. In this context, AR augments the physical world with digital overlays, while VR places users in fully simulated environments. This paper discusses these technologies in detail, including their basic concepts and hardware and software components. This survey examines how AR and VR can positively impact artistic fields such as virtual art galleries, augmented public installations, and innovative theatrical performances. We discuss limitations in hardware, software development, user experience, and ethical considerations. Further, we emphasize collaboration possibilities, accessibility, and inclusivity to probe AR and VR's profound impact on artistic creativity. The paper illustrates the transformative power of these technologies through case studies and noteworthy projects. Finally, future trends are outlined, highlighting advancements, emerging artistic forms, and social and cultural implications.

Author 1: Fan Wang

Author 2: Zonghai Zhang

Author 3: Liangyi Li

Author 4: Siyu Long

Keywords: Virtual reality; augmented reality; artistic expression; emerging technologies; immersive experiences

Paper 66: NovSRC: A Novelty-Oriented Scientific Collaborators Recommendation Model

Abstract: Collaborator recommendation is a crucial topic in research management. This paper proposes a Novelty-Oriented Scientific Research Collaborator recommendation model (NovSRC). By recommending collaborators under the guidance of novel indicators, NovSRC aims to broaden scholars' research perspectives and facilitate the progress of research innovation. NovSRC utilizes heterogeneous academic networks composed of different academic entities and their relationships to learn vector representations of scholars and quantify their novelty metrics. A weighted academic collaboration network was constructed by measuring the novelty collaboration strength (NCS) among scholars under the novelty index, and based on this network, the final vector representation of scholars under the guidance of novelty characteristics was learned. By calculating the similarity between scholar vectors, NovSRC generates a Top-N recommendation list with a focus on novelty. The experimental results indicate that NovSRC achieved the best recommendation performance. Compared with the baseline models, the recommendation precision of NovSRC has improved by 6.9%, the F1 value has increased by 17.3%, and the novelty collaboration strength among scholars has increased by 3.3%. The analysis of the recommended list shows that compared to the target scholars, scholars recommended by the NovSRC model exhibit a wider distribution of research interests, which confirms that novelty has become a key benchmark factor for scholars seeking collaborators.

Author 1: Xiuxiu Li

Author 2: Mingyang Wang

Author 3: Chaoran Wang

Author 4: Yujia Fu

Author 5: Xianjie Wang

Keywords: Scientific collaborator recommendation; novelty; heterogeneous academic collaboration network; network representation learning

Paper 67: A Deep Learning Model for Prediction of Cardiovascular Disease Using Heart Sound

Abstract: Cardiovascular disease is the most emerging disease in this generation of youth. You need to know about your heart condition to overcome this disease appropriately. An electronic stethoscope is used in the cardiac auscultation technique to listen to and analyze heart sounds. Several pathologic cardiac diseases can be detected by auscultation of the heart sounds. Unlike heart murmurs, the sounds of the heart are separate; brief auditory phenomena usually originate from a single source. This article proposes a deep-learning model for predicting cardiovascular disease. The combined deep learning model uses the MFCC and LSTM for feature extraction and prediction of cardiovascular disease. The model achieved an accuracy of 94.3%. The sound dataset used in this work is retrieved from the UC Irvine Machine Learning Repository. The main focus of this research is to create an automated system that can assist doctors in identifying normal and abnormal heart sounds.

Author 1: Rohit Ravi

Author 2: P. Madhavan

Keywords: Cardiovascular disease; prediction; LSTM; MFCC; deep learning

Paper 68: Optimized Deep Belief Networks Based Categorization of Type 2 Diabetes using Tabu Search Optimization

Abstract: Diabetics mellitus has the potential to result in numerous challenges. Based on the increasing morbidity rates in recent years, it is projected that the global diabetic population will surpass 642 million by 2040, indicating that approximately one in every ten individuals will have diabetes. Undoubtedly, this alarming statistic necessitates urgent focus from both academics as well as industry to foster novelty and advancement in prediction of diabetics, with the aim of preserving patients' lives. Deep learning (DL) was employed to forecast a multitude of ailments as a result of its swift advancement. Nevertheless, DL approaches continue to face challenges in achieving optimal prediction performance as a result of the selection of hyper-parameters and tuning of parameters. Hence, the careful choice of hyper-parameters plays a crucial role in enhancing classification performance. This paper introduces TSO-DBN, a Tabu Search Optimization method (TSO) that is based on Deep Belief Network (DBN). TSO-DBN has demonstrated exceptional performance in several medical fields. The Tabu Search Optimization algorithm (TSO) has been used to pick hyper-parameters and optimize parameters. During the experiment, two problems were tackled in order to improve the findings. The TSO-DBN model exhibited exceptional performance, surpassing other models with an accuracy of 96.23%, an F1-score of 0.8749, and a Matthews Correlation Coefficient (MCC) of 0.88.63.

Author 1: Smita Panigrahy

Author 2: Sachikanta Dash

Author 3: Sasmita Padhy

Keywords: Deep belief network; Tabu search; diabetics mellitus; hyper-parameters; optimization

Paper 69: A Model for Automatic Code Generation from High Fidelity Graphical User Interface Mockups using Deep Learning Techniques

Abstract: Graphical user interface (GUI) is the most prevalent type of user interfaces (UI) due to its visual nature, which allows direct manipulation and interaction with the software. Mockup-based design is a frequently used workflow for constructing GUI. In this workflow, the anticipated UI design process typically progresses through multiple steps, culminating in the creation of a higher fidelity mockup and subsequent implementation of that mockup into code. The design process involves repeating those multiple steps because of the ongoing changes in requirements, which can make the process tedious and necessitate modifications to the GUI code. Additionally, the process of implementing and converting a design into GUI code itself is laborious and time-consuming task that can prevent developers from dedicating the bulk of their time implementing the software's functionality and logic, making it a costly endeavor. Automating the code generation process using GUI design images can be a solution to mitigate these issues and allow more time to be allocated towards building the application's functionality. In this research paper, deep learning object detectors are employed to detect the predominant UI elements and their spatial arrangement in a high-fidelity UI mockup image. This approach generates an intermediate representation, including the layout hierarchy of the user interface leading to the automation of the front-end code generation process for the mockup. The proposed approach demonstrates its effectiveness through experimental results, achieving a recognition mean average precision (mAP) of 91.37% for atomic elements and 87.40% for container elements in the mockup image. Additionally, similarity metrics are employed to assess the visual resemblance between the generated mockups and the original ones.

Author 1: Michel Samir

Author 2: Ahmed Elsayed

Author 3: Mohamed I. Marie

Keywords: Code generation; graphical user interfaces; deep learning; computer vision; mockups

Paper 70: NTDA: The Mitigation of Denial of Service (DoS) Cyberattack Based on Network Traffic Detection Approach

Abstract: Security is one of the important aspects which is used to protect data availability from being compromised. Denial of service (DoS) attack is a common type of cyberattack and becomes serious security threats to information systems and current computer networks. DoS aims to explicit attempts that will consume and disrupt victim resources to limit access to information services by flooding a target system with a high volume of traffic, thereby preventing the availability of the resources to the legitimate users. However, several solutions were developed to overcome the DoS attack, but still suffer from limitations such as requiring additional hardware, fail to provide a unified solution and incur a high delay of detection accuracy. Therefore, the network traffic detection approach (NTDA) is proposed to detect the DoS attack in a more optimistic manner based on various scenarios. First, the high network traffic measurements and mean deviation, second scenario relied on the transmission rate per second (TPS) of the sender. The proposed algorithm NTDA was simulated using MATLAB R2020a. The performance metrics taken into consideration are false negative rate, accuracy, detection rate and true positive rate. The simulation results show that the performance parameters of proposed NTDA algorithm outperformed in DoS detection the other well-known algorithms.

Author 1: Muhannad Tahboush

Author 2: Adel Hamdan

Author 3: Firas Alzobi

Author 4: Moath Husni

Author 5: Mohammad Adawy

Keywords: Network security; DoS attack; cyberattack; network traffic

Paper 71: Word2vec-based Latent Semantic Indexing (Word2Vec-LSI) for Contextual Analysis in Job-Matching Application

Abstract: Job-matching applications have become a technology that provides solutions for making decisions about accepting and looking for work. The contextual analysis of documents or data from job matching is needed to make decisions. Some existing studies on the analysis of job-matching applications can use the Latent Semantic Indexing (LSI) method, which is based on word-to-word comparisons in the text. LSI has the advantage of contextual analysis. It can analyze amounts of data above 10,000 words. However, the conventional LSI method has limitations in contextual analysis because it uses the exact words but different meanings. Therefore, this paper proposes a new technique called word2vec-based latent semantic indexing (Word2vec-LSI) for contextual analysis, which is based on gen-sim as a multi-language word library. Then, modeling in text and wordnet and stopword as basic text modeling. We then used word2vec-LSI to perform contextual analysis based on the Irish (IE), Swedish (SE), and United Kingdom (UK) languages in the dataset (Jobs on CareerBuilder UK). The results of applying conventional LSI have an accuracy level of 79%, recall has a value of 79%, precision has a value of 62%, and Fi-Scor has a value of 70% with a similarity level of up to 50%. After implementing word2vec-LSI, it can increase accuracy, recall, and precision, and Fi-Scor both have 84% in contextual analysis, and the similarity level reaches up to 95%. Experiments confirm the usefulness of word2vec-LSI in increasing accuracy for contextual analysis applicable in natural language text mining.

Author 1: Sukri Sukri

Author 2: Noor Azah Samsudin

Author 3: Ezak Fadzrin

Author 4: Shamsul Kamal Ahmad Khalid

Author 5: Liza Trisnawati

Keywords: Contextual; LSI; job-matching; text-base; word2vec

Paper 72: An Effective Forecasting Approach of Temperature Enabling Climate Change Analysis in Saudi Arabia

Abstract: Climate change is a global issue with far-reaching consequences, and understanding regional temperature patterns is critical for effective climate change analysis. In this context, accurate forecasting of temperature is critical for mitigating and understanding its impact. This study proposes an effective temperature forecasting approach in Saudi Arabia, a region highly vulnerable to climate change's effects, particularly rising temperatures. The approach uses advanced neural networks models such as the Long Short-Term Memory (LSTM), Gate Recurrent Unit (GRU), and Bidirectional LSTM (BiLSTM) model. A comparative analysis of these models is also introduced to determine the most effective model for forecasting the mean values of temperatures in the following years, understanding climate variability, and informing sustainable adaptation strategies. Several experiments are conducted to train and evaluate the models on a time series data of temperatures in Saudi Arabia, taken from a public dataset of countries' historical global average land temperatures. Performance metrics such as Mean Absolute Error (MAE), Mean Relative Error (MRE), Root Mean Squared Error (RMSE), and coefficient of determination (R-squared) are employed to measure the accuracy and reliability of each model. Experimental results show the models' ability to capture short-term fluctuations and long-term trends in temperature patterns. The findings contribute to the advancement of climate modeling methodologies and offer a basis for selecting a suitable model in similar environmental contexts.

Author 1: Sultan Noman Qasem

Author 2: Samah M. Alzanin

Keywords: Climate change; Saudi Arabia; temperature; forecasting; recurrent neural network models

Paper 73: Utilizing the Metaverse in Astrosociology: Examine Students' Perspectives of Space Science Education

Abstract: Big economic countries must invest in space skills to create a favorable business environment, particularly in KSA considering the present mindset in outer space. KSA's vast landmass is a tremendous asset that makes it the perfect position to provide space services throughout the Middle East and the world. Space science education is becoming increasingly important, requiring advanced technology and computational skills to benefit early-career scientists. The Ministry of Education in KSA has declared that students will take Earth and Space Sciences to prepare them for global competition. Traditional learning experiences seem to have little to no impact on students' conceptual understandings of the space science courses. The sociological interests of Generation Z serve as the foundation for modern Metaverse approaches. Students' comprehension and interest in studying space and the galaxy are increased by provided a simulation of space travel using metaverse technology. The major goal of this study is to underline the significance and usefulness of employing metaverse technology while creating a new space science curriculum to advance knowledge in the field of space scientific education. Another goal is to introduce the value of astrosociology in understanding how people might interact with one another in space. A voluntary survey was completed by 39 students prior to their training in the metaverse space simulation as part of this study. They then used the space simulation with careful observation. After that, they reply to a follow-up survey. The findings supported the suggestion that the metaverse should be included in space science curricula. A number of comments and interests also arise on the viability of space travel, social interaction, and the advantages of using the metaverse to research these issues.

Author 1: Yahya Almurtadha

Keywords: Metaverse; space science education; astrosociology; virtual reality; space simulation

Paper 74: Student Outcome Assessment on Structured Query Language using Rubrics and Automated Feedback Generation

Abstract: Automated assessment of student assignment based on SQL(Structured Query Language) queries is an efficient method for evaluating and providing feedback on their DBMS-related skills. This paper provides a three step approach of how student submissions are assessed automatically using various machine learning approaches and introduced an automated grading system for SQL(Structured Query Language) queries. ASQGS (Automated SQL Query Grading System) is the process of evaluating SQL queries submitted by students of a classroom. Due to the difficulties involved in the automatic grading procedure, this endeavor continues to attract the researcher's interest in developing a new and superior grading system. The purpose of this study is to demonstrate how text relevance is calculated between a reference query that the teacher sets and a query that the student submits. To compute the grade, the similarity value between the student and reference queries will be compared. In this paper various feature similarity techniques were discussed which is required before applying the machine learning model to automatically assess the grade of the student’s SQL assignment. In the second step the grade received by the ASQG is used for student outcome assessment using rubrics with respect to Bloom’s taxonomy and finally scores can be calculated using predefined rubrics criteria. Additionally, in the 3rd step the system can generate feedback for students, highlighting specific areas of improvement, errors, or suggestions to enhance their queries among different groups of students segregated by their SQL knowledge.

Author 1: Sidhidatri Nayak

Author 2: Reshu Agarwal

Author 3: Sunil Kumar Khatri

Author 4: Masoud Mohammadian

Keywords: Automated SQL Query grading system; Cosine similarity; LSA; Multinomialnb; KNN; Logistic regression; student outcome assessment; rubric; feedback

Paper 75: Genetic Algorithms and Feature Selection for Improving the Classification Performance in Healthcare

Abstract: Microarray technology appeared recently and is used in genetic research to study gene expressions. Microarray has been widely applied to many fields, especially the health sector, such as diagnosing and predicting diseases, specifically cancer diseases. These experiments usually generate a huge amount of gene expression data with analytical and computational complexities. Therefore, feature selection techniques and different classifications help solve these problems by eliminating irrelevant and redundant features. This paper presents a proposed method for classifying the data using eight classifications machine learning algorithms. Then, the Genetic Algorithm (GA) is applied to improve the selection of the best features and parameters for the model. We use the higher accuracy of the model among the different classifications as a measure of fit in the genetic algorithm; this means that the model’s accuracy can be used to select the best solutions than others in the community. The proposed method was applied to the colon, breast, prostate, and Central Nervous System (CNS) diseases and experimental outcomes demonstrated an accuracy rate of 93.75, 96.15, 82.76, and 93.33 respectively. Based on these findings, the proposed method works well and effectively.

Author 1: Alaa Alassaf

Author 2: Eman Alarbeed

Author 3: Ghady Alrasheed

Author 4: Abdulsalam Almirdasie

Author 5: Shahd Almutairi

Author 6: Mohammed Abullah Al-Hagery

Author 7: Faisal Saeed

Keywords: Cancer classification; gene expression; feature selection; microarray data; algorithm; machine learning; genetic algorithm

Paper 76: Optimization Strategy for Industrial Machinery Product Selection Scheme Based on DMOEA

Abstract: With the continuous innovation and replacement of industrial machinery products, the traditional optional configuration plans are no longer able to complete product selection work with high quality. To further optimize the product selection process and solve the multi-objective selection problem of industrial machinery products, a multi-objective problem model for product selection is normalized and constructed based on the existing difficulties in industrial machinery product selection. A new product selection model is proposed by introducing a multi-objective evolutionary algorithm based on density calculation for model solving. The experimental results showed that the new model had the highest selection success rate of 97% and selection accuracy close to 95% when the iterations were 250. In addition, the maximum absolute error sum of the selected bearing and bearing seat diameters under this model was 0.002. The maximum relative error was 0.01%. The highest reliability of algorithm fitting was 99.9%. Simulation tests found that the average selection success rate was 93%. The average selection quality loss was 26%. In summary, the new selection model proposed in the study has certain advantages and feasibility. It can provide effective decision-making solutions for the design and selection of industrial machinery products.

Author 1: Shichang Liu

Author 2: Xinbin Yang

Author 3: Haihua Huang

Keywords: Industrial machinery products; optional configuration plan; multi objective evolutionary algorithm; density calculation; selection success rate

Paper 77: A Deep Learning Framework for Detection and Classification of Implant Manufacturer using X-Ray Radiographs

Abstract: Now-a-days, artificial prosthesis is widely used to mitigate pain in damaged shoulders and restore their movement ability. The process involves a complex surgery that attempts to fix an artificial prosthesis into a dead shoulder as a replacement for the ball and socket joints of the shoulder. Long after the surgical process, the need for revision or reoperation may arise due to some problems with the prosthesis. Identification of prosthesis manufacturer is a paramount step in the reoperation exercise. Traditional approach compares the prosthesis under consideration with prosthesis from a vast number of manufacturers. This approach is cost-efficient and requires no extra training for the physician to identify the prosthesis manufacturer. However, the method is time inefficient and is prone to mistakes. Systems based on machine learning have the potential to reduce human errors and expedite the revision process. This paper proposes a shallow 2D convolution neural network (CNN) for the classification of shoulder prosthesis To speed-up the learning process and improve the performance of the deep learning model for implant classification, this paper employed three different techniques. Firstly, a generative adversarial network (GAN) is applied to the dataset to augment the classes with fewer samples to ensure the data imbalance problem is eliminated. Secondly, the highly discriminating features are extracted using principal component analysis (PCA) and used to train the proposed model. Lastly, the model hyper-parameters are optimised to ensure optimal model performance. The model trained with extracted features with a variance of 0.99 achieved the best accuracy of 99.8%.

Author 1: Attar Mahay Sheetal

Author 2: K. Sreekumar

Keywords: Machine learning; deep learning; convolution neural network; Adversarial Network (GAN); Principal Component Analysis (PCA); shoulder implants

Paper 78: Rolling Bearing Life Prediction Technology Based on Feature Screening and LSTM Model

Abstract: As one of the important components of industrial equipment, the health condition of rolling bearings will directly affect the operational effectiveness of the equipment. Therefore, to ensure equipment safety and reduce maintenance costs, an intelligent rolling bearing life prediction technology is proposed. Firstly, it extracts the fault information of rolling bearings and introduces Fisher score for feature selection. Simultaneously, a variational modal analysis method on the grounds of improved particle swarm optimization is introduced to achieve denoising of rolling bearing signals. Finally, an improved bidirectional long short-term model is introduced to construct a prediction model and achieve the life prediction of rolling bearings. In the performance analysis of the denoising model, the optimal modal component K value of the denoising model was obtained through experimental analysis as 3, and the optimal penalty factor number was 1000. In the time-domain signal analysis of the two models, the proposed model possesses a more excellent decomposition effect on the original signal compared to the comparative model, and the signal denoising ability is improved by 26.35%. In the prediction of rolling bearing life, the proposed model can accurately predict the early and late life of rolling bearings. For example, when the collection time is 100, the actual remaining life is 0.712, and the proposed model is 0.721, which is better than other models. In the comparison of average absolute error, the proposed model is 0.035, which outperforms other models. This indicates that the proposed rolling bearing life prediction model has excellent predictive performance. The research provides essential technical references for the maintenance of industrial machinery and equipment, as well as equipment life monitoring.

Author 1: Yujun Zhao

Keywords: Features; rolling bearings; prediction; fisher score; bidirectional long short term model

Paper 79: Retrieval-Augmented Generation Approach: Document Question Answering using Large Language Model

Abstract: This study introduces the Retrieval Augmented Generation (RAG) method to improve Question-Answering (QA) systems by addressing document processing in Natural Language Processing problems. It represents the latest breakthrough in applying RAG to document question and answer applications, overcoming previous QA system obstacles. RAG combines search techniques in vector store and text generation mechanism developed by Large Language Models, offering a time-efficient alternative to manual reading limitations. The research evaluates RAG's that use Generative Pre-trained Transformer 3.5 or GPT-3.5-turbo from the ChatGPT model and its impact on document data processing, comparing it with other applications. This research also provides datasets to test the capabilities of the QA document system. The proposed dataset and Stanford Question Answering Dataset (SQuAD) are used for performance testing. The study contributes theoretically by advancing methodologies and knowledge representation, supporting benchmarking in research communities. Results highlight RAG's superiority: achieving a precision of 0.74 in Recall-Oriented Understudy for Gisting Evaluation (ROUGE) testing, outperforming others at 0.5; obtaining an F1 score of 0.88 in BERTScore, surpassing other QA apps at 0.81; attaining a precision of 0.28 in Bilingual Evaluation Understudy (BLEU) testing, surpassing others with a precision of 0.09; and scoring 0.33 in Jaccard Similarity, outshining others at 0.04. These findings underscore RAG's efficiency and competitiveness, promising a positive impact on various industrial sectors through advanced Artificial Intelligence (AI) technology.

Author 1: Kurnia Muludi

Author 2: Kaira Milani Fitria

Author 3: Joko Triloka

Author 4: Sutedi

Keywords: Natural Language Processing; Large Language Model; Retrieval Augmented Generation; Question Answering; GPT

Paper 80: A User Control Framework for Cloud Data Migration in Software as a Service

Abstract: Cloud computing represents the overarching paradigm that enables organizations to leverage cloud services for data storage and application deployment. Nowadays, organizations that use the cloud services can migrate their data using software as a service (SaaS). The organizations’ data and application are deployed over the cloud through the cloud data migration process of the on-premise to cloud migration; referring to the transition process from the legacy, locally hosted systems to cloud environment. Several data migration frameworks have emerged to guide users in the migration process. While numerous studies have addressed the importance of granting control to users during the cloud data migration process, a user control framework is yet to be created. Thereby, depriving user of visibility and sense of ownership, customization to meet users need, compliance and governance, and training. This paper aims at achieving this by proposing a conceptual user control framework for cloud data migration process in SaaS. The framework is constructed based on a comprehensive analysis conducted over existing research works that are related to cloud data migration with the aim to identify the steps/phases of data migration process, the factors affecting the user control with regard to the identified phases, and the control metrics of each identified factor. An initial conceptual user control framework is constructed based on the analysis of the literatures and further enhancement of the framework is made based on the expert reviews.

Author 1: Danga Imbaji Injuwe

Author 2: Hamidah Ibrahim

Author 3: Fatimah Sidi

Author 4: Iskandar Ishak

Keywords: Comparative analysis; cloud computing; cloud data migration; on-premises to cloud migration; user control; Software as a Service

Paper 81: Blockchain-Enabled Cybersecurity Framework for Safeguarding Patient Data in Medical Informatics

Abstract: Securing patient information is crucial in the quickly changing field of healthcare informatics to guarantee privacy, reliability, and adherence to legal requirements. This article presents a complete cybersecurity architecture enabled by blockchain and customized for the medical informatics area. The framework initiatives to provide adequate safeguards for sensitive patient data by utilizing AES-Diffie-Hellman key exchange for secure communication, blockchain technology with Proof-of-Work (PoW), and Role-Based Access Control (RBAC) for fine access management. A strong cybersecurity architecture is crucial for maintaining the security, credibility, and availability of private patient information in the current healthcare information management environment. By using decentralized storage, access control methods, and cutting-edge encryption strategies, the suggested framework overcomes these difficulties. The framework ensures safe data transport and storage by showcasing effective AES encryption as well as decryption procedures through performance evaluation. PoW consensus combined with blockchain technology provides the framework with auditable and immutable data storage, reducing the possibility of data manipulation and unwanted access. Additionally, granular access control is made possible by the integration of RBAC, guaranteeing that only those with the proper authorization may access patient data. Python is used to implement the suggested framework. The suggested method considerably outperformed NTRU, RSA, and DES with encryption and decryption times of 12.1 and 12.2 seconds, respectively. The proposed Blockchain-Enabled Cybersecurity Framework demonstrates exceptional efficacy, evidenced by its ability to achieve a 97.9% reduction in unauthorized access incidents, thus offering robust protection for patient data in medical informatics.

Author 1: Prajakta U. Waghe

Author 2: A Suresh Kumar

Author 3: Arun B Prasad

Author 4: Vuda Sreenivasa Rao

Author 5: E. Thenmozhi

Author 6: Sanjiv Rao Godla

Author 7: Yousef A.Baker El-Ebiary

Keywords: Block Chain; Cybersecurity; Diffie Hellmen; Patient Data; Proof of Work

Paper 82: Reliable Hybridization Approach for Estimation of The Heating Load of Residential Buildings

Abstract: In recent times, the world's growing population, coupled with its ever-increasing energy demands, has led to a significant rise in the consumption of fossil fuels. Consequently, this surge in fossil fuel usage has exacerbated the threat of global warming. Building energy consumption represents a significant portion of global energy usage. Accurately determining the energy consumption of buildings is crucial for effective energy management and preventing excessive usage. In pursuit of this goal, this study introduces a novel and robust machine learning (ML) method based on the K-nearest Neighbor (KNN) algorithm for predicting the heating load of residential buildings. While the KNN model demonstrates satisfactory performance in predicting heating loads, for the attainment of optimal results and accuracy, two novel optimizers, the Snake Optimizer (SO) and the Black Widow Optimizer (BWO), have been incorporated into the hybridization of the KNN model. The results highlight the effectiveness of KNSO in predicting heating load, as evidenced by its impressive R2 value of 0.986 and the low RMSE value of 1.231. This breakthrough contributes significantly to the ever-pressing pursuit of energy efficiency in the built environment and its pivotal role in addressing global environmental challenges.

Author 1: Huanhuan Li

Keywords: Heating load; residential buildings; k-nearest neighbor; snake optimizer; black widow optimizers

Paper 83: Enhancing Security in IoT Networks: Advancements in Key Exchange, User Authentication, and Data Integrity Mechanisms

Abstract: Future Internet (FI) will be shaped by the Internet of Things (IoT), however because of their limited resources and varied communication capabilities, IoT devices present substantial challenges when it comes to securing connectivity. The adoption of robust security measures is hindered by limited compute power, memory, and energy resources, hence diminishing the promise for improved IoT capabilities. Confidentiality, integrity, and authenticity are ensured via authentication mechanisms are influenced by privacy needs, which are driven by sorts of customers that IoT networks service. Authentication is crucial in vital industries like linked cars and smart cities where hackers might use holes to access sensor data. Verification of the Gate Way Node (GWN), which is responsible for mutual authentication, user and sensor registration, and session key creation, is essential. The efficiency of key creation has been enhanced to tackle temporal intricacies linked to different key lengths. With notable advantages, the novel method shortens the time required to generate cryptographic keys: only 60 milliseconds for 100-bit keys and 120 milliseconds for 256-bit keys. This improvement fortifies resistance against new cyber threats by strengthening security basis of IoT networks and enhancing responsiveness and dependability. Through open transmission channels, users send login requests, and after successfully authenticating, they create session keys to establish secure connections with cloud servers. Python simulation results show how resilient the system is to security threats while preserving affordable interaction, processing, and storage. This development not only strengthens IoT networks but also guarantees their sustainability in the face of changing security threats.

Author 1: Alumuru Mahesh Reddy

Author 2: M. Kameswara Rao

Keywords: IoT; Public key; key authentication; gate way node; data integrity mechanisms

Paper 84: Student Performance Estimation Through Innovative Classification Techniques in Education

Abstract: In the current era of intense educational competition, institutions must effectively classify individuals based on their abilities, proactively forecast student performance, and work towards enhancing their forthcoming examination outcomes. Providing early guidance to students is crucial in helping them focus their efforts on specific areas to boost their academic achievements. This analytical approach supports educational institutions in mitigating failure rates by utilizing students' previous performance in relevant courses to predict their outcomes in a specific program. Data mining encompasses a variety of techniques used to reveal hidden patterns within vast datasets. In the context of educational data mining, these methods are applied within the educational sphere, with a specific emphasis on analyzing data from both students and educators. These patterns can offer significant value for predictive and analytical objectives. In this study, Gaussian Process Classification (GPC) was employed for the prediction of student performance. To improve the model's accuracy, two cutting-edge optimizers, namely the Golden Eagle Optimizer (GEO) and the Pelican Optimization Algorithm (POA), were incorporated. When assessing the model's performance, four widely used metrics were utilized: Accuracy, Precision, Recall, and F1-score. The results of this study underscore the effectiveness of both the POA and GEO optimizers in enhancing GPC performance. Specifically, GPC+GEO demonstrated remarkable effectiveness in the Poor grade, while GPC+POA excelled in the Acceptable and Excellent category. This highlights the positive impact of these optimization techniques on the model's predictive capabilities.

Author 1: Hui Fan

Author 2: Guoping Zhu

Author 3: Jianhua Zhan

Keywords: Student performance; Gaussian Process Classification; Golden Eagle Optimizer; Pelican Optimization Algorithm

Paper 85: Penetration Testing Framework using the Q Learning Ensemble Deep CNN Discriminator Framework

Abstract: Penetration testing (PT) serves as an effective tool for examining networks and identifying vulnerabilities by simulating a hacker's attack to uncover valuable information, such as details about the host's operating and database systems. Strong penetration testing is crucial for assessing system vulnerabilities in the constantly changing world of cyber security. Existing methods often struggle with adapting to dynamic threats, providing limited automation, and lacking the ability to discern subtle security weaknesses. In comparison to manual PT, intelligent PT has gained widespread popularity due to its efficiency, resulting in reduced time consumption and lower labor costs. Considering this, the effective penetration testing framework is developed using prairie natural swarm (PNS) optimized Q-learning ensemble deep CNN. Initially, the penetration testing environment (Shodan search engine) is simulated, and along with that expert knowledge base is also generated. Subsequently, the Nmap script engine and Metasploit are deployed, providing robust tools for network investigation and vulnerability assessment. The system state is then relayed to the Q-learning ensemble deep convolutional neural network (Q-learning ensemble deep CNN) classifier. This unique ensemble combines the strengths of Q-learning and deep CNNs, enabling optimal policy learning for decision-making. The prairie natural swarm optimization algorithm is developed through the hybridization of coyote and particle swarm characteristics to fine-tune classifier parameters, enhancing performance. Additionally, the discriminator is trained to maximize standard action rewards while minimizing discounted action rewards, distinguishing valuable from less valuable information. By evaluating the advantage function, successful penetration likelihood is determined, informing situational decision-making through the Q-learning ensemble deep CNN classifier. Accuracy, sensitivity, and specificity as well as the proposed PNS-optimized Q-learning ensemble deep model are used to evaluate the output. In comparison to other approaches currently in use, CNN achieves values of 94.54%, 94.40%, 94.90% for TP, 94.64%, 94.69%, and 94.52% for k-fold.

Author 1: Dipali Nilesh Railkar

Author 2: Shubhalaxmi Joshi

Keywords: Penetration testing; Q-learning; ensemble deep CNN; prairie natural swarm optimization; Nmap script engine

Paper 86: Educational Data Mining in European Union – Achievements and Challenges: A Systematic Literature Review

Abstract: The quality of education is one of the pillars of sustainable development, as set out in “The 2030 Agenda for Sustainable Development”, adopted by all United Nations Member States in 2015. Recent social and technological developments, as well as events such as the COVID-19 pandemic or conflicts in many parts of the world, have led to essential changes in the way education processes are carried out. In addition, they have made it possible to generate, collect and store large amounts of data related to these processes, data that can hide useful information for decisions that, in the medium or long term, can lead to a significant increase in the quality of education. Uncovering this information is the subject of Educational Data Mining. To understand the state-of-the-art reflected by recent developments, trends, theories, methodologies, and applications in this field, in the European Union, we considered it appropriate to conduct a systematic and critical literature review. Our paper aims to identify, analyze, and synthesize relevant information from these articles, both to build a foundation for further studies and to identify gaps or unexplored issues that can be addressed in future research. The analysis is based on research identified in three international databases recognized for content quality: Scopus, Science direct, and IEEEXplore.

Author 1: Corina Simionescu

Author 2: Mirela Danubianu

Author 3: Bogdanel Constantin Gradinaru

Author 4: Marius Silviu Maciuca

Keywords: Educational data mining; systematic literature review; European Union; Kitchenham methodology; data mining techniques

Paper 87: Enhancing Cryptojacking Detection Through Hybrid Black Widow Optimization and Generative Adversarial Networks

Abstract: Cybercriminals now find cryptocurrency mining to be a lucrative endeavour. This is frequently seen in the form of cryptojacking, which is the illegal use of computer resources for cryptocurrency mining. Protecting user resources and preserving the integrity of digital ecosystems depend heavily on the detection and mitigation of such threats. This research presents a unique method that combines Black Widow Optimisation (HBWO) with Generative Adversarial Networks (GANs) to improve the detection of cryptojacking. Due to its covert nature and tendency to elude conventional detection methods, cryptojacking is still a widespread concern. In order to overcome this difficulty, our work makes use of the complementary abilities of deep learning and metaheuristic optimisation. To maximise feature selection for efficient identification of cryptojacking activity, BWO—which draws inspiration from the foraging behaviour of spiders—is utilised. Simultaneously, GANs are employed to produce artificial intelligence (AI) augmentations, which strengthen the detection model's resilience and enrich the training dataset. Utilising HBWO to identify the most discriminative features is the first step in our technique, which also includes preprocessing the dataset to extract pertinent features. The training dataset is then supplemented with artificial data samples created using GANs, which enhances the detection model's capacity for generalisation. Experiments conducted on real-world datasets show the effectiveness of our solution, outperforming baseline techniques. The hybrid technique that has been suggested offers a viable way to improve the detection of cryptojacking. Through the combination of HBWO for feature optimisation and GANs for data augmentation, our approach demonstrates improved 98.02% accuracy and resilience in detecting cryptojacking activity. With its novel framework for fending against new dangers in the digital sphere, this research adds to the continuing efforts in cybersecurity.

Author 1: Meenal R. Kale

Author 2: Deepa

Author 3: Anil Kumar N

Author 4: N. Lakshmipathi Anantha

Author 5: Vuda Sreenivasa Rao

Author 6: Sanjiv Rao Godla

Author 7: E. Thenmozhi

Keywords: Cryptojacking; attack detection; Generative Adversarial Networks; Black Widow Optimization; cybercriminals

Paper 88: DeepEmoVision: Unveiling Emotion Dynamics in Video Through Deep Learning Algorithms

Abstract: Emotion detection from videos plays a pivotal role in understanding human behavior and interaction. This study delves into a cutting-edge method that leverages Recurrent Neural Networks (RNN), Support Vector Machines (SVM), K-Nearest Neighbours (KNN), Convolutional Neural Networks (CNN) and to precisely detect emotions exhibited in video content, holding significant importance in comprehending human behavior and interactions. The devised approach entails a multi-phase procedure: initially, employing CNN-based feature extraction to isolate facial expressions and pertinent visual cues by extracting and pre-processing video frames. These extracted features capture intricate patterns and spatial information crucial for discerning emotions. The results of the trials show that CNN, SVM, KNN, and RNN have promising performance, highlighting their potential. Among the other machine learning models, RNN has attained a 95% accuracy rate in recognizing and classifying emotions in video information. This combination of approaches provides a thorough plan for identifying emotions in dynamic visual material in real time.

Author 1: Prathwini

Author 2: Prathyakshini

Keywords: Emotion detection; video analysis; Recurrent Neural Networks (RNN); Support Vector Machines (SVM); K-Nearest Neighbours (KNN); Convolutional Neural Networks (CNN); facial expression recognition; machine learning

Paper 89: Botnet Detection and Incident Response in Security Operation Center (SOC): A Proposed Framework

Abstract: In the dynamic landscape of evolving cyber threats, Security Operations Centers (SOCs) play an important role in protecting digital assets. Among these threats, botnets are particularly challenging due to their ability to take over many devices and launch coordinated attacks. Through comparative analysis, the research gaps in existing frameworks have been identified. Based on these insights, a botnet detection and incident response framework aligned with SOC practices has been proposed. This proposed framework emphasizes proactive measures by integrating threat intelligence, detection and monitoring tools to detect botnet attack and facilitate rapid response. Future research will focus on conducting evaluation and validation studies to assess the effectiveness and performance of the framework in controlled environments. This effort will contribute to develop the framework and ensuring it aligns with practical cybersecurity needs.

Author 1: Roslaily Muhammad

Author 2: Saiful Adli Ismail

Author 3: Noor Hafizah Hassan

Keywords: Botnet detection; threat incident response; security operation center

Paper 90: Enhancing Customer Segmentation Insights by using RFM + Discount Proportion Model with Clustering Algorithms

Abstract: In this digital era, the use of e-commerce has expanded and is widely adopted by society. One of the reasons why people use e-commerce platforms is because of their convenience and ease of use. However, the rapid growth of e-commerce has led to a substantial rise in transactions within the platform, involving various business entities. Therefore, it is crucial to perform customer segmentation to group them based on their purchasing behavior. The implementation of data mining techniques, such as clustering, is highly beneficial in this case. Clustering helps process datasets and transform them into useful information. In this study, transaction data obtained from one of the e-commerce stores, i.e. MurahJaya888 and followed by analysis using various clustering methods such as K-means, K-medoids, Fuzzy c-means, and Mini-batch k-means. We also proposed a new model that will become the attributes cluster, namely, RFM + DP (Discount Proportion). The Discount Proportion Rate will provide more insights for customer segmentation as it helps understand purchasing behavior that is more responsive to discount utilization. Implementing these four clustering methods with RFM + DP model resulted in four clusters based on the optimal elbow method. Furthermore, the evaluation and performance metrics for each clustering algorithm indicate that Mini Batch K-Means achieved the highest silhouette score of 0.50. Meanwhile, K-Means obtained the highest CH index value compared to the other algorithms, which was 1056.

Author 1: Victor Hugo Antonius

Author 2: Devi Fitrianah

Keywords: Clustering; RFM; discount proportion; customer segmentation; data mining

Paper 91: Adapting Outperformer from Topic Modeling Methods for Topic Extraction and Analysis: The Case of Afaan Oromo, Amharic, and Tigrigna Facebook Text Comments

Abstract: Facebook users generate a vast amount of data, including posts, comments, and replies, in various formats such as short text, long text, structured, unstructured, and semi structured. Consequently, obtaining import information from social media data becomes a significant challenge for low-resource languages such as Afaan Oromo, Amharic, and Tigrigna. Topic modeling algorithms are designed to identify and categorize topics within a set of documents based on their semantic similarity which helps obtain insight from documents. This study proposes latent Dirichlet allocation, matrix factorization, probabilistic latent semantic analysis, and BERTopic to extract topics from Facebook text comments in Afaan Oromo, Amharic, and Tigrigna. The study utilized text comments from the Facebook pages of various individuals, including activists, politicians, athletes, media companies, and government offices. BERTopic was found to be the most effective for identifying major topics and providing valuable insights into user conversations and social media trends with coherence scores of 82.74%, 87.85%, and 81.79% respectively.

Author 1: Naol Bakala Defersha

Author 2: Kula Kekeba Tune

Author 3: Solomon Teferra Abate

Keywords: Afaan oromo; amharic; tigrigna; BERTopic; topic extraction; social media data

Paper 92: Artificial Intelligence System for Malaria Diagnosis

Abstract: Malaria threats have remained one of the major global health issues over the past decades specifically in low-middle income countries. 70% of the Kenya population lives in malaria endemic zones and the majority have barriers to access health services due to factors including lack of income, distance, and social culture. Despite various research efforts using blood smears under a microscope to combat malaria with advantages, this method is time consuming and needs skillful personnel. To effectively solve this issue, this study introduces a new method integrating InfoGainAttributeEval feature selection techniques and parameter tuning method based on Artificial Intelligence and Machine Learning (AIML) classifiers with features to diagnose types of malaria more accurately. The proposed method uses 100 features extracted from 4000 samples. Sets of experiments were conducted using Artificial Neural Network (ANNs), Naïve Bayes (NB), Random Forest (RF) classifiers and Ensemble methods (Meta Bagging, Random Committee Meta, and Voting). Naïve Bayes has the best result. It achieved 100% accuracy and built the model in 0.01 second. The results demonstrate that the proposed method can classify malaria types accurately and has the best result compared to the reported results in the field.

Author 1: Phoebe A Barracloug

Author 2: Charles M Were

Author 3: Hilda Mwangakala

Author 4: Gerhard Fehringer

Author 5: Dornald O Ohanya

Author 6: Harison Agola

Author 7: Philip Nandi

Keywords: Malaria diagnosis; malaria symptoms; artificial intelligence and machine learning classifier; malaria classifier

Paper 93: A New Time-Series Classification Approach for Human Activity Recognition with Data Augmentation

Abstract: Accurate classification of multivariate time series data represents a major challenge for scientists and practitioners exploring time series data in different domains. LSTM-Auto-encoders are Deep Learning models that aim to represent input data efficiently while minimizing information loss during the reconstruction phase. Although they are commonly used for Dimensionality Reduction and Data Augmentation, their potential in extracting dynamic features and temporal patterns for temporal data classification is not fully exploited in contrast to the tasks of time-series prediction and anomaly detection. In this article, we present a multi-level hybrid TSC-LSTM-Auto-Encoder architecture that takes full advantage of the incorporation of temporal labels to capture comprehensively temporal features and patterns. This approach aims to improve the performance of temporal data classification using this additional information. We evaluated the proposed architecture for Human activity Recognition (HAR) using the UCI-HAR and WISDM public benchmark datasets. The achieved performance outperforms the current state-of-the-art methods.

Author 1: Youssef Errafik

Author 2: Younes Dhassi

Author 3: Adil Kenzi

Keywords: Deep Learning (DL); multivariate time series; Time Series Classification (TSC); Human Activity Recognition (HAR)

Paper 94: Adaptive Threshold Tuning-based Load Balancing (ATTLB) for Cost Minimization in Cloud Computing

Abstract: Cloud computing has revolutionized the on-demand resource provisioning through virtualization. However, dynamic pricing of cloud resources presents cost management challenges. Load balancing is critical for cloud efficiency; however, current algorithms use static thresholds and are unable to adapt to fluctuating prices. This study proposes a novel Dynamic Threshold Tuning (ATTLB) algorithm that optimizes the CPU and memory thresholds of a load balancer based on real-time pricing. The ATTLB algorithm has a pricing monitor to track spot prices; a VM profiler to record capacities; a threshold optimizer to tune thresholds based on pricing, capacity, and SLAs; and a load dispatcher to assign requests to VMs using the optimized thresholds. Extensive simulations compare ATTLB with weighted round-robin (WRR), ant colony optimization (ACO), and least connection-based load balancing (LCLB) algorithms using the CloudSim toolkit. The results demonstrate the ability of ATTLB to reduce total costs by over 35% and improve SLA violations by 41% compared with prior techniques for cloud load balancing. Adaptive threshold tuning provides robustness against dynamic pricing and demand changes. ATTLB balances cost, performance, and utilization through real-time threshold adaptation.

Author 1: Lama S. Khoshaim

Keywords: Cloud computing; load balancing; threshold optimization; cost minimization; pricing models; CloudSim; resource allocation; cost-aware load balancing

Paper 95: Facial Emotion Recognition-based Engagement De-tection in Autism Spectrum Disorder

Abstract: Engagement is the state of alertness that a person experiences and the deliberate focus of their attention on a task-relevant stimulus. It positively correlates with many aspects such as learning, social support, and acceptance. Facial emotion recog-nition using artificial intelligence can be beneficial to automati-cally measure individual engagement especially when using au-tomated learning and playing modalities such as using Robots. In this study, we proposed an automatic engagement detection model through facial emotional recognition, particularly in de-termining autistic children’s engagement. The methodology em-ployed a transfer learning approach at the dataset level, utiliz-ing facial image datasets from typically developing (TD) chil-dren and children with ASD. The classification task was per-formed using convolutional neural network (CNN) methods. Comparative analysis revealed that the CNN method demon-strated superior accuracy compared to random forest (RF), support vector machine (SVM), and decision tree algorithms in both the TD and ASD datasets. The findings highlight the poten-tial of CNN-based facial emotion recognition for accurately assessing engagement in children with ASD, with implications for enhancing learning, social support, and acceptance in this population. This research contributes to the field of engagement measurement in autism and underscores the importance of lev-eraging AI techniques for improving understanding and support for children with ASD.

Author 1: Noura Alhakbani

Keywords: Engagement detection; facial emotion recognition; autistic chil-dren; convolutional neural networks

Paper 96: State-Feedback Control of Ball-Plate System: Geometric Approach

Abstract: This research focuses on investigating the issue of accurately controlling the location of the ball in the ball and plate system. The findings of this research have practical applications across several domains, including optimizing the alignment of solar panels to enhance their energy generation capacity. In this work, we propose the development of a system dynamics model using the Euler-Lagrangian approach. Furthermore, we analyze a technique in the frequency domain known as the geometric approach to create a state-feedback control that ensures the stability of the system. This study primarily focuses on analyzing the characteristic equations associated with the closed-loop system, while also considering the impact of feedback delay. Ultimately, the proposed technique is substantiated by presenting simulation data for validation.

Author 1: Khalid Lefrouni

Author 2: Saoudi Taibi

Keywords: Ball-plate system; Delay systems; Geometric approach; State-feedback control

Paper 97: Maximizing Solar Panel Efficiency in Partial Shade: The Improved POA Solution for MPPT

Abstract: This paper presents an innovative approach to improving Maximum Power Point Tracking (MPPT) in solar photovoltaic (PV) systems affected by partial shading, a common challenge that significantly reduces efficiency. Our research focuses on enhancing the Pelican Optimization Algorithm (POA), a promising tool in solar energy optimization, to better tackle the efficiency drop observed under shaded conditions. The enhancements to the POA involve the integration of advanced adaptive mechanisms that enable more precise response to the fluctuating irradiance patterns typical of partially shaded environments. This revised version of the POA demonstrates remarkable adaptability and precision in identifying and tracking the maximum power point, significantly outperforming its original iteration. The methodology of this study encompasses a series of rigorous simulations and real-world testing scenarios, designed to evaluate the POA's performance under various degrees and patterns of shading. The results show a notable improvement in efficiency, with the enhanced POA maintaining high levels of energy capture even in suboptimal sunlight conditions. Additionally, the improved algorithm exhibits robustness against the rapid changes in irradiance, which is characteristic of partially shaded solar PV systems. Our findings underscore the potential of the enhanced POA as a robust, adaptive solution for optimizing solar energy collection, offering significant benefits for solar installations in geographies prone to shading. This work not only contributes to the field of renewable energy optimization but also provides valuable insights for the development of more resilient and efficient solar energy systems.

Author 1: Youssef Mhanni

Author 2: Youssef Lagmich

Keywords: Pelican Optimization Algorithm (POA); Maximum Power Point Tracking (MPPT); Solar Photovoltaic Systems; Partial Shading

Paper 98: An Integrated CNN-BiLSTM Approach for Facial Expressions

Abstract: Deep learning algorithms have demonstrated good performance in many sectors and applications. Facial expression recognition (FER) is recognizing the emotions through images. FER is an integral part of many applications. With the help of the CNN-BiLSTM integrated approach, higher accuracy can be achieved in identification of the facial expressions. Convolutional neural networks (CNN) consist of a Conv2D layer, dividing the given images into batches, performing normalization and if required flattening the data i.e. converting the data in a 1D array and achieving a higher accuracy. BiLSTM works on two LSTMs i.e. one in the forward direction and the other in a backward direction. One can use LSTM to process the images (datasets) however, it is suggested with the help of BiLSTM can predict the expressions with more accuracy. Input data is available in both the direction (forward and backward) which helps maintaining the context. Using LSTM CNN and BiLSTM always helps increasing the prediction accuracy. Application areas where a BiLSTM can give more prediction accuracy are the forecasting models, text recognition, speech recognition, classifying the large data and the proposed facial expression recognition. The integrated approach (CNN and BiLSTM) increases the accuracy significantly as discussed in the results and discussion section. This approach could be categorized as a fusion technique where two methods (approaches) are integrated to get higher accuracy. The results and discussion section elaborates the effectiveness of the integrated approach compared to HERO: human emotions recognition for realizing the intelligent internet of things. As compared to the HERO approach CNN-BiLSTM gives good results in terms of precision and recall.

Author 1: B. H. Pansambal

Author 2: A. B. Nandgaokar

Author 3: J. L. Rajput

Author 4: Abhay Wagh

Keywords: CNN (Convolutional Neural Network); BiLSTM (Bi Directional Long Short Term Memory); facial expression recognition; deep learning; flattening

Paper 99: Research on Innovative Design of Towable Caravans Integrating Kano-AHP and TRIZ Theories

Abstract: The caravan industry in China is facing significant challenges, primarily because the mode of caravan travel is relatively niche within the country and the industry as a whole has had a slow start. This has ultimately resulted in a mismatch between the design aesthetics of caravans and the preferences of Chinese consumers. Based on the foundation of understanding user preferences, this study proposes a new design methodology that integrates the Kano model, the Analytic Hierarchy Process (AHP), and TRIZ theory to align with the preferences of Chinese users. Initially, a Kano model is constructed based on the suggestions from experts and users to categorize user needs. Subsequently, the AHP method is employed to reclassify the key needs identified in the Kano model, establish judgment matrices, and develop a scoring system to provide a scientific basis for design decisions. Finally, TRIZ theory is applied to address potential physical and technical contradictions encountered during the design process, thereby developing practical and aesthetically pleasing caravan design solutions.

Author 1: Jinyang Xu

Author 2: Xuedong Zhang

Author 3: Aihu Liao

Author 4: Shun Yu

Author 5: Yanming Chen

Author 6: Longping Chen

Keywords: Kano model; towed caravans; exterior design; Analytic Hierarchy Process (AHP); TRIZ theory

Paper 100: Enhancing Employee Performance Management

Abstract: Human resource management (HRM) plays a crucial role in the effective functioning of modern businesses. However, As the volume of data continues to increase, HR professionals are facing growing challenges in objectively gathering, measuring, and interpreting human resources data. The research problem addressed in this study is the need to improve methods for the objective classification of teams based on the most relevant performance factors considering the subjectivity of current tools. To tackle this issue, the research questions focus on the possibility of developing an efficient model for team classification using supervised machine learning algorithms. This study consists of developing and validating three team classification models using the support vector machine (SVM), the K-nearest neighbor (KNN) algorithm, and the multiple linear regression algorithm (MLR) after using PCA for data reduction. Following extensive validation, the module based on MLR was identified as the most effective, achieving an accuracy of 87.5% in Predicting employee performance, which makes it possible to anticipate and fill employee skills gaps and optimize recruiting efforts. This work provides human resources professionals with a data-driven decision support to enhance Human Resources Management using Machine Learning.

Author 1: Zbakh Mourad

Author 2: Aknin Noura

Author 3: Chrayah Mohamed

Author 4: Bouzidi Abdelhamid

Keywords: HRM; HR analytics; Employee Performance Prediction; Support Vector Machine (SVM) Algorithm; K-Nearest Neighbor (KNN) Algorithm; Multiple Linear Regression (MLR) algorithm; Principal Component Analysis (PCA)

Paper 101: Advancing Strawberry Disease Detection in Agriculture: A Transfer Learning Approach with YOLOv5 Algorithm

Abstract: Strawberry Disease Detection in the Agricultural Sector is of paramount importance, as it directly impacts crop yield and quality. A multitude of methods have been explored in the literature to address this challenge, but deep learning techniques have consistently demonstrated superior accuracy in disease detection. Nevertheless, the current research challenge in deep learning-based strawberry disease detection remains the demand for consistently high accuracy rates. In this study, we propose a deep learning model based on the Yolov5 architecture to address the aforementioned research challenge effectively. Our approach involves the generation of a custom dataset tailored to strawberry disease detection and the execution of comprehensive training, validation, and testing processes to fine-tune the model. Experimental results and performance evaluations were conducted to validate our proposed method, demonstrating its ability to achieve accurate results consistently. This research contributes to the ongoing efforts to enhance strawberry disease detection methods within the agricultural sector, ultimately aiding in the early identification and mitigation of diseases to preserve crop yield and quality.

Author 1: Chunmao LIU

Keywords: Strawberry disease detection; deep learning; agricultural; YOLOv5 model; training

Paper 102: Profiling and Classification of Users Through a Customer Feedback-based Machine Learning Model

Abstract: The systems aimed at predicting user preferences and providing recommendations are now commonly used in many systems such as online shops, social websites, and tourist guide websites. These systems typically rely on collecting user data and learning from it in order to improve their performance. In the context of urban mobility, user Profiling and Classification represent a crucial step in the continuous enhancement of services provided by our multi-agent system for multimodal transportation. In this paper, our goal is to implement and compare some machine learning (ML) algorithms. We will address the technical aspect of this implementation, demonstrating this model leverages customer feedback to develop a thorough understanding of individual preferences and travel behaviors. Through this approach, we can categorize users into distinct groups, enabling a finer personalization of route recommendations and transportation preferences. The ML model analyzes customer feedback, identifies recurring patterns, and continuously adjusts user profiles based on their evolution. This innovative approach aims to optimize the user experience by offering more precise and tailored recommendations, while fostering dynamic adaptation of the system to the changing needs of urban users.

Author 1: Jihane LARIOUI

Author 2: Abdeltif EL BYED

Keywords: Machine learning; urban mobility; multimodal transportation; multi-agent systems

Paper 103: Detection of Harassment Toward Women in Twitter During Pandemic Based on Machine Learning

Abstract: Harassment is an offensive behavior, intimidating and could cause discomfort to the victims. In some cases, the harassments could lead to a traumatic experience to the vulnerable victims. Currently, the harassments towards women in social media have become more daring and are rising. The increasing number of the social media users since the Covid-19 pandemic in 2020 might be one of the factor. Due to the problem, this research aims to assist in detecting the harassment sentiments toward women in Twitter. The sentiment analysis is based on a machine learning approach and Support Vector Machine (SVM) has been chosen due its acceptable performance in sentiment classification. The objective of the research is to explore the capability of SVM in the detection of harassments toward women in Twitter. The research methodology covers the data collection using Tweepy, data preprocessing, data labelling using TextBlob, feature extraction using TF-IDF vectorizer and dataset splitting using the Hold-Out method. The algorithm was evaluated using the Confusion Matrix and the ROC analysis. The algorithm was integrated with the Graphical User Interface (GUI) using Streamlit for ease of use. The implementation of the SVM algorithm in detecting the harassments toward women was successful and reliable as it achieved good performance, with 81% accuracy. The recommendations for the SVM model improvement is to train the dataset of other languages and to collect the Twitter data regularly. The performance of SVM would also be compared with other machine learning algorithms for further validations.

Author 1: Wan Nor Asyikin Wan Mustapha

Author 2: Norlina Mohd Sabri

Author 3: Nor Azila Awang Abu Bakar

Author 4: Nik Marsyahariani Nik Daud

Author 5: Azilawati Azizan

Keywords: Harassment; women; detection; twitter; SVM

Paper 104: Deep Learning-based Food Calorie Estimation Method in Dietary Assessment: An Advanced Approach using Convolutional Neural Networks

Abstract: Dietary pattern assessments, essential for chronic illness management and well-being, involve time-consuming manual data input and food intake remembering. A more dependable and automated approach is needed since such procedures may create mistakes and inconsistencies. This study solves a long-standing problem by automating nutritional assessment using deep learning and image analysis. CNNs, deep learning models for image processing, were employed in our study. Food category algorithms are trained with thousands of pictures. Even with numerous food items, these models can distinguish them in digital photographs. Our method calculates food portions after identification. Photometric food measurements are obtained using reference objects like plates and forks. Yet another deep learning model predicts portions. The method evaluates food calories last. Select food types and portions are matched to nutritional databases. These findings might automate, enhance, and user-centrically assess food intake in health informatics. Our first experiments are encouraging, but we must understand the approach's limits and need for refinement. The findings underpin future research and development. This approach envisions a future where patients can monitor their nutrition and doctors can get accurate data. This may prevent and treat lifestyle problems.

Author 1: Kalivaraprasad B

Author 2: Prasad M.V.D

Author 3: Naveen kishore Gattim

Keywords: Deep learning; convolutional neural networks; food calorie estimation; dietary assessment; computer vision; health informatics

Paper 105: Multi-Objective Reinforcement Learning for Virtual Machines Placement in Cloud Computing

Abstract: The rapid demand for cloud services has provoked cloud providers to efficiently resolve the problem of Virtual Machines Placement in the cloud. This paper presents a VM Placement using Reinforcement Learning that aims to provide optimal resource and energy management for cloud data centers. Reinforcement Learning provides better decision-making as it solves the complexity of VM Placement problem caused due to tradeoff among the objectives and hence is useful for mapping requested VM on the minimum number of Physical Machines. An enhanced Tournament-based selection strategy along with Roulette Wheel sampling has been applied to ensure that the optimization goes through balanced exploration and exploitation, thereby giving better solution quality. Two heuristics have been used for the ordering of VM, considering the impact of CPU and memory utilizations over the VM placement. Moreover, the concept of the Pareto approximate set has been considered to ensure that both objectives are prioritized according to the perspective of the users. The proposed technique has been implemented on MATLAB 2020b. Simulation analysis showed that the VMRL performed preferably well and has shown improvement of 17%, 20% and 18% in terms of energy consumption, resource utilization and fragmentation respectively in comparison to other multi-objective algorithms.

Author 1: Chayan Bhatt

Author 2: Sunita Singhal

Keywords: Virtual machines placement; cloud computing; reinforcement learning; energy consumption; resource utilization

Paper 106: A New Aerial Image Segmentation Approach with Statistical Multimodal Markov Fields

Abstract: Aerial images, captured by drones, satellites, or aircraft, are omnipresent in diverse fields, from mapping and surveillance to precision agriculture. The efficacy of image analysis in these domains hinges on the quality of segmentation, and the precise delineation of objects and regions of interest. In this context, leveraging Markov fields for aerial image segmentation emerges as a promising avenue. The segmentation of aerial images presents a formidable challenge due to the variability in capture conditions, lighting, vegetation, and environmental factors. To meet this challenge, the work proposes an innovative method harnessing the power of Markov fields by integrating a multimodal energy function. This energy function amalgamates key attributes, including color difference measured by the CIEDE2000 metric, texture features, and detected edge information. The CIEDE2000 metric, derived from the CIELab color space, is renowned for its ability to measure color difference more consistently with human perception than conventional metrics. By incorporating this metric into the energy function, the approach enhances sensitivity to subtle color variations crucial for aerial image segmentation. Texture, a vital attribute characterizing regions in aerial images, offers crucial insights into terrain or objects. The method incorporates texture features to refine the separation of homogeneous regions. Contours, playing a fundamental role in segmentation, are identified using an edge detector to pinpoint boundaries between regions of interest. This information is integrated into the energy function, elevating contour consistency and segmentation accuracy. This article comprehensively presents the methodological approach, the conducted experiments, obtained results, and a thorough discussion of the method's advantages and limitations.

Author 1: Jamal Bouchti

Author 2: Ahmed Bendahmane

Author 3: Adel Asselman

Keywords: Image segmentation; multimodal markov fields statistical integration; CIEDE2000 color difference; texture features; edge information

Paper 107: Path Planning and Control of Intelligent Delivery UAV Based on Internet of Things and Edge Computing

Abstract: This paper investigates the intelligent delivery UAV path planning and control problem based on the Internet of Things and edge computing, and proposes a novel model and algorithm to realize the collaborative optimization of the path planning and control of the UAV, which improves the intelligence level and flight efficiency of the UAV. In this paper, the mathematical model of UAV path planning and control is firstly established, the relationship and influencing factors among the elements of UAV, edge server, delivery task, path planning and control are analyzed, and the optimization objectives and constraints are proposed. Then, this paper designs an algorithmic framework for UAV path planning and control, using the support and guidance of edge computing to achieve the cooperative optimization of path planning and control of UAVs, taking into account the constraints and objectives of the UAVs themselves, as well as the synergy and competition between UAVs. Then, this paper proposes specific algorithms for UAV path planning and control, adopting methods such as meta-heuristics, to solve the optimization problem of UAV path planning and control, and improve the intelligent level and flight performance of UAVs.

Author 1: Xiuzhu Zhang

Keywords: Internet of things; edge computing; smart distribution; drone path; planning and control

Paper 108: Secure IoT Seed-based Matrix Key Generator

Abstract: The rapid evolution of the Internet of Things (IoT) has significantly transformed various aspects of both personal and professional spheres, offering innovative solutions in fields from home automation to industrial manufacturing. This progression is driven by the integration of physical devices with digital networks, facilitating efficient communication and data processing. However, such advancements bring forth critical security challenges, especially regarding data privacy and network integrity. Conventional cryptographic methods often fall short in addressing the unique requirements of IoT environments, such as limited device computational power and the need for efficient energy consumption. This paper introduces a novel approach to IoT security, inspired by the principles of steganography – the art of concealing information within other non-secret data. This method enhances security by embedding secret information within the payload or communication protocols, aligning with the low-power and minimal processing capabilities of IoT devices. We propose a steganographic key generation algorithm, adapted from the Diffie-Hellman key exchange model, tailored for IoT. This approach eliminates the need for explicit parameter exchange, thereby reducing vulnerability to key interception and unauthorized access, prevalent in IoT networks. The algorithm utilizes a pre-shared 2D matrix and a synchronized seed-based approach for covert communication without explicit data exchange. Furthermore, we have rigorously tested our algorithm using the NIST Statistical Test Suite (STS), comparing its execution time with other algorithms. The results underscore our algorithm's superior performance and suitability for IoT applications, highlighting its potential to secure IoT networks effectively without compromising on efficiency and device resource constraints. This paper presents the design, implementation, and potential implications of this algorithm for enhancing IoT security, ensuring the full realization of IoT benefits without compromising user security and privacy.

Author 1: Youssef NOUR-EL AINE

Author 2: Cherkaoui LEGHRIS

Keywords: Security; IoT; steganography; key exchange; cryptography

Paper 109: Hierarchical Spatiotemporal Aspect-Based Sentiment Analysis for Chain Restaurants using Machine Learning

Abstract: In recent years, aspect-based sentiment analysis of restaurant business reviews has emerged as a pivotal area of research in natural language processing (NLP), aiming to provide detailed analytical methods benefiting both consumers and industry professionals. This study introduces a novel approach, Hierarchical Spatiotemporal Aspect-Based Sentiment Analysis (HISABSA), which combines lexicon-based methods such as VADER Lexicon, the AFFIN model, and TextBlob with contextual methods. By integrating advanced machine learning (ML) techniques, this hybrid methodology facilitates sentiment analysis, empowering chain restaurants to assess changes in sentiments towards specific aspects of their services across different branches and over time. Leveraging transformer-based models such as RoBERTa and BERT, this approach achieves effective sentiment classification and aspect extraction from text reviews. The results demonstrate the reliability of extracting valid aspects from online reviews of specific branches, offering valuable insights to business owners striving to succeed in competitive markets.

Author 1: Mouyassir Kawtar

Author 2: Abderrahmane Fathi

Author 3: Noureddine Assad

Author 4: Ali Kartit

Keywords: HISABSA; hybrid model; NLP; ML; VADER Lexicon; AFFIN model; TextBlob; ABSA; Restaurant reviews; Transformer-based models; Lexicon-based methods; RoBERTa model; BERT model

Paper 110: Enhancing Water Quality Forecasting Reliability Through Optimal Parameterization of Neuro-Fuzzy Models via Tunicate Swarm Optimization

Abstract: Forecasting water quality is critical to environmental management because it facilitates quick decision-making and resource allocation. On the opposite hand, current methods are not always able to produce reliable forecasts, which is often due to challenges in parameter optimization for complex models. This research presents a novel approach to enhance the forecasting accuracy of water quality by optimizing neuro-fuzzy models using Tunicate Swarm Optimisation (TSO). The introduction highlights the limitations of current techniques as well as the necessity for precise estimates of water quality. One of the drawbacks is that neuro-fuzzy models are not well-modelled, which makes it harder for them to identify the minute patterns in data on water quality. The suggested approach is unique in that it applies TSO, an optimization algorithm inspired by nature that emulates tunicates' behaviour, to the neuro-fuzzy models' parameter optimization process. The highly complex parameter space is effectively navigated by TSO's swarm intelligence, which strikes a balance between exploration and exploitation to improve model performance. To optimize model parameters, the process comprises three steps: creating an objective function, defining the neuro-fuzzy model, and seamlessly integrating TSO. By mimicking the motions of tunicates as they look for the best conditions in the marine environment, TSO constantly optimizes the variables. Experiments demonstrate that the proposed strategy is more effective than traditional optimization techniques in forecasting water quality. As seen by the optimised neuro-fuzzy model's increased prediction accuracy and several dataset validations, Tunicate Swarm Optimisation has potential for reliable environmental forecasting. This work presents a potential path for improved environmental decision-making systems by offering an optimisation strategy inspired by nature that overcomes the limitations of existing methods and enhances water quality forecasting tools.

Author 1: Kambala Vijaya Kumar

Author 2: Y Dileep Kumar

Author 3: Sanjiv Rao Godla

Author 4: Mohammed Saleh Al Ansari

Author 5: Yousef A.Baker El-Ebiary

Author 6: Elangovan Muniyandy

Keywords: Water quality forecasting; neuro-fuzzy models; tunicate swarm optimization; parameter optimization; environmental decision support

Paper 111: Revolutionizing Healthcare by Unleashing the Power of Machine Learning in Diagnosis and Treatment

Abstract: Machine learning (ML) is a versatile technology that has the potential to revolutionize various industries. ML can predict future trends in customer expectations that allow organizations to develop new products accordingly. ML is a crucial field of data science that uses different algorithms to predict insights and improve decision-making. The widespread acceptance of ML algorithms ML can provide helpful information using the enormous volume of health data generated regularly. Quicker diagnoses by doctors can be delivered by adopting ML techniques that can bring down medical charges and applying pattern identification algorithms to examine medical images. Every technology brings its challenges; in the same way, ML also has several challenges in healthcare that need to be acknowledged before we witness complete automation in medical diagnosis. People are still forbidden to share their personal information with intermediaries for treatment. Medical record governance is essential to ensure that health records are not missed. Manual diagnosis often goes in the wrong direction, as doctors are also human. Lack of communication between medical workers and patients, considering the insufficient data to diagnose disease, sometimes results in deteriorating health conditions. This paper deals with an introduction to machine learning. These ML algorithms are widely used for health diagnosis, a comparison analysis of literature work that has been done so far, existing challenges of the healthcare system, healthcare industry using machine learning applications, real-life use cases, practical implementation of disease prediction, and conclusion with its future scope.

Author 1: Medini Gupta

Author 2: Sarvesh Tanwar

Author 3: Salil Bharany

Author 4: Faisal Binzagr

Author 5: Hadia Abdelgader Osman

Author 6: Ashraf Osman Ibrahim

Author 7: Samsul Ariffin Abdul Karim

Keywords: Machine Learning; Health Diagnosis; Supervised Learning; Prediction; Classification

Paper 112: Ceramic Microscope Image Classification Based on Multi-Scale Fusion Bottleneck Structure and Chunking Attention Mechanism

Abstract: In recent years, the status of ceramics in fields such as art, culture, and historical research has been continuously improving. However, the increase in malicious counterfeiting and forgery of ceramics has disrupted the normal order of the ceramic market and brought challenges to the identification of authenticity. Due to the intricate and interfered nature of the microscopic characteristics of ceramics, traditional identification methods have been suffering from issues of low accuracy and efficiency. To address these issues, there is a proposal for a multi-scale fusion bottleneck structure and a chunking attention module to improve the neural network model of Resnet50 and perform ceramic microscopic image classification and recognition. Firstly, the original bottleneck structure has been replaced with a multi-scale fusion bottleneck structure, which can establish a feature pyramid and establish associations between different feature layers, effectively focusing on features at different scales. Then, chunking attention modules are added to both the shallow and deep networks, respectively, to establish remote dependencies in low-level detail features and high-level semantic features, to reduce the impact of convolutional receptive field restrictions. The experimental results show that, in terms of classification accuracy and other indicators, this model surpasses the mainstream neural network models with a better classification accuracy of 3.98%compared to the benchmark model Resnet50, achieving 98.74%. Meanwhile, in comparison with non-convolutional network models, it has been found that convolutional models are more suitable for the recognition of ceramic microscopic features.

Author 1: Zhihuang Zhuang

Author 2: Xing Xu

Author 3: Xuewen Xia

Author 4: Yuanxiang Li

Author 5: Yinglong Zhang

Keywords: Deep learning; ceramic anti-counterfeiting; image classification; attention mechanism

Paper 113: Weighted PSO Ensemble using Diversity of CNN Classifiers and Color Space for Endoscopy Image Classification

Abstract: Endoscopic image is a manifestation of visualization technology to the human gastrointestinal tract, allowing detection of abnormalities, characterization of lesions, and guidance for therapeutic interventions. Accurate and reliable classification of endoscopy images remains challenging due to variations in image quality, diverse anatomical structures, and subtle abnormalities such as polyps and ulcers. Convolutional Neural Network (CNN) is widely used in modern medical imaging, especially for abnormality classification tasks. However, relying on a single CNN classifier limits the model’s ability to capture endoscopy images’ full complexity and variability. A potential solution to the problem involves employing ensemble learning, which combines multiple models to reach at a final decision. Nevertheless, this learning approach presents several challenges, notably a significant risk of data bias. This issue arises from the unequal influence of weak and strong learners in most ensemble strategies, such as standard voting, which usually depend on certain assumptions, including equal performance among the models. However, it reduces the capability towards diverse model collaboration. Therefore, this paper proposes two solutions to the problems. Firstly, we create a diverse pool of CNNs with end-to-end approach. This approach promotes model diversity and enhances confidence in making a final decision. Secondly, we propose employing Particle Swarm Optimization to enhance the weight of the members in the ensemble learner in order to create a more resilient and accurate model compared to the standard ensemble learning approach. The experiment demonstrates that the proposed ensemble model outperforms the baseline model on both the Kvasir 1 and Kvasir 2 datasets, highlighting the effectiveness of the suggested approach in integrating diverse information from the baseline model. This enhanced performance highlights the efficacy of capturing diverse information from the baseline model.

Author 1: Diah Arianti

Author 2: Azizi Abdullah

Author 3: Shahnorbanun Sahran

Author 4: Wong Zhyqin

Keywords: Convolution neural network; particle swarm optimization; diversity; weighted ensemble

Paper 114: Research Octane Number Prediction Based on Feature Selection and Multi-Model Fusion

Abstract: The catalytic cracking-based process for lightening heavy oil yields gasoline products with sulfur and olefin contents surpassing 95%, consequently diminishing the Research Octane Number (RON) of gasoline during desulfurization and olefin reduction stages. Hence, investigating methodologies to mitigate RON loss in gasoline while maintaining effective desulfurization is imperative. This study addresses this challenge by initially performing data cleaning and augmentation, employing box plot modeling and Grubbs’ test for outlier detection and removal. Subsequently, through the integration of mutual information and the Lasso method, data dimensionality is reduced, with the top 30 variables selected as primary factors. A predictive model for RON loss is then established based on these 30 variables, utilizing Random forest and Support Vector Regression (SVR) models. Employing this model enables the computation of RON loss for each data sample. Comparing with existing methods, our approach could ensure a balance between effective desulfurization and mitigated RON loss in gasoline products.

Author 1: Junlin Gu

Keywords: Feature selection; random forest model; support vector machine model; RON loss

Paper 115: Spherical Fuzzy Z-Numbers-based CRITIC CRADIAS and MARCOS Approaches for Evaluating English Teacher Performance

Abstract: Consider combining quantitative and qualitative data for these case studies, such as interviews with English teachers, student evaluations, classroom observations, and surveys. Contextual elements, including community support, resources, and school demographics, should also be taken into consideration. The assessment process in English teaching performance evaluation is very complicated and diverse, making it a perfect fit for use in the Multi-Attribute Group Decision Making (MAGDM) framework. The utilization of Spherical Fuzzy Z˘-Number Sets (SF Z˘NS) is essential in Multi-Attribute Group Decision Making (MAGDM) to handle intricate problems. These sets are significantly more capable of handling higher levels of uncertainty than the fuzzy set designs used today. Here, we provide a method, Compromise Ranking of Alternatives from Distance to Ideal Solution (CRADIAS), designed to address MAGDM problems in SF Z˘NS, particularly in cases when attribute weights are opaque. Attribute weights may be found by applying the CRITIC technique. The first section of the research covers the examination of spherical fuzzy Z numbers, their accuracy and scoring functions, and the main concept behind their functioning. We then propose the use of spherical fuzzy Z˘-Number data to handle MAGDM cases in a decision-making process. This work strengthens the topic’s theoretical underpinnings as well as its practical applicability. By conducting a comparison study, we apply the MARCOS approach to validate and illustrate the validity of our findings. This methodical approach guarantees a thorough evaluation of the suggested method’s effectiveness and adds to the current discussion on how to make wise decisions in difficult and uncertain situations.

Author 1: Jie Niu

Keywords: SFZNS; CRITRIC technique; CRADIAS method; MARCOS method

Paper 116: Issuance Policies of Route Origin Authorization with a Single Prefix and Multiple Prefixes: A Comparative Analysis

Abstract: Resource Public Key Infrastructure (RPKI) is a solution to mitigate the security issues faced by inter-domain routing. Within the RPKI framework, Route Origin Authorization (ROA) plays a crucial role as an RPKI object. ROA allows address space holders to place a single IP address prefix or multiple IP address prefixes in it. However, this feature has introduced security risks during the global deployment of RPKI. In this study, we analyze the current status of ROA issuance and discuss the impact of using two ROA issuance policies on RPKI security and synchronization efficiency. Based on the aforementioned work, recommendations are proposed for the utilization of ROA issuance policies.

Author 1: Zetong Lai

Author 2: Zhiwei Yan

Author 3: Guanggang Geng

Author 4: Hidenori Nakazato

Keywords: BGP; RPKI; route origin authorization; inter-domain routing security; computer network protocols; routing

Paper 117: Scientometric Analysis and Knowledge Mapping of Cybersecurity

Abstract: Cybersecurity research includes several areas, such as authentication, software and hardware vulnerabilities, and defences against cyberattacks. However, only a limited number of cybersecurity experts have a comprehensive understanding of all aspects of this sector. Hence, it is vital to possess an impartial comprehension of the prevailing patterns in cyberse-curity research. Scientometric analysis and knowledge mapping may effectively detect cybersecurity research trends, significant studies, and emerging technologies within this particular context. The main aim of this research is to comprehend the developmental trend of the academic literature about the concepts of “malware detection” and ‘cybersecurity’. We collected 9,967 publications from January 2019 to December 2023 and used the Citespace tool for scientometric analysis. This study found six co-citation clusters,namely malware classification, evading malware classifier, android malware detection, IoT network, CNN, and ransomeware families. Additionally, this study discovered that the top contributing countries are the USA, China, and India based on the citation count, and the Chinese Academy of Science, the University of California, and the University of Texas are the top contributing institutions based on the frequency of the publications.

Author 1: Fahad Alqurashi

Author 2: Istiak Ahmad

Keywords: Cybersecurity; cyber threats; scientometric analysis; bibliomatic analysis

Paper 118: SCEditor: A Graphical Editor Prototype for Smart Contract Design and Development

Abstract: In recent years, particularly with the Ethereum blockchain’s advent, smart contracts have gained significant interest as a means of regulating exchanges among multiple parties via code. This surge has prompted the emergence of various smart contract (SC) programming languages, each possessing distinct philosophies, grammatical structures, and components. Conse-quently, developers are increasingly involved in SC programming. However, these languages are platform specific, implying that a transition to another platform necessitates the use of different languages. Additionally, developers require a certain level of control over SCs to address encountered bugs and ensure maintenance. To address these developer-centric challenges, this paper presents SCEditor, a novel Eclipse Sirius-based prototype editor designed for the visualization, design, and creation of SCs. The editor proposes a means of standardizing the usage of SC programming languages through the incorporation of graphical syntax and a metamodel conforming to Model-Driven Engineering (MDE) principles and SC construction rules to generate an abstract SC model. The efficacy of this editor is demonstrated through testing on a voting SC written in Vyper and Solidity languages. Furthermore, the editor holds potential for future exploitation in model transformation and code generation for various SC languages.

Author 1: Yassine Ait Hsain

Author 2: Naziha Laaz

Author 3: Samir Mbarki

Keywords: Blockchain; Metamodel; Model-driven Engineering (MDE); Smart Contract (SC); SC Programming

Paper 119: Software Defect Prediction via Generative Adversarial Networks and Pre-Trained Model

Abstract: Software defect prediction, which aims to predict defective modules during software development, has been implemented to assist developers in identifying defects and ensure software quality. Traditional defect prediction methods utilize manually designed features such as “Lines Of Code” that fail to capture the syntactic and semantic structures of code. Moreover, the high cost and difficulty of building the training set lead to insufficient data, which poses a significant challenge for training deep learning models, particularly for new projects. To overcome the practical challenge of data limitation and improve predictive capacity, this paper presents DP-GANPT, a novel defect prediction model that integrates generative adversarial networks and state-of-the-art code pre-trained models, employing a novel bi-modal code-prompt input representation. The proposed approach explores the use of code pre-trained model as auto-encoders and employs generative adversarial networks algorithms and semi-supervised learning techniques for optimization. To facilitate effective training and evaluation, a new software defect prediction dataset is constructed based on the existing PROMISE dataset and its associated engineering files. Extensive experiments are performed on both within-project and cross-project defect prediction tasks to evaluate the effectiveness of DP-GANPT. The results reveal that DP-GANPT outperforms all the state-of-the-art baselines, and achieves performance comparable to them with significantly less labeled data.

Author 1: Wei Song

Author 2: Lu Gan

Author 3: Tie Bao

Keywords: Software defect prediction; semi-supervised learning; generative adversarial networks; deep learning

Paper 120: Enhancing Model Robustness and Accuracy Against Adversarial Attacks via Adversarial Input Training

Abstract: Adversarial attacks present a formidable challenge to the integrity of Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) models, particularly in the domain of power quality disturbance (PQD) classification, necessitating the development of effective defense mechanisms. These attacks, characterized by their subtlety, can significantly degrade the performance of models critical for maintaining power system stability and efficiency. This study introduces the concept of adversarial attacks on CNN-LSTM models and emphasizes the critical need for robust defenses.We propose Input Adversarial Training (IAT) as a novel defense strategy aimed at enhancing the resilience of CNN-LSTM models. IAT involves training models on a blend of clean and adversarially perturbed inputs, intending to improve their robustness. The effectiveness of IAT is assessed through a series of comparisons with established defense mech-anisms, employing metrics such as accuracy, precision, recall, and F1-score on both unperturbed and adversarially modified datasets.The results are compelling: models defended with IAT exhibit remarkable improvements in robustness against adver-sarial attacks. Specifically, IAT-enhanced models demonstrated an increase in accuracy on adversarially perturbed data to 85%, a precision improvement to 86%, a recall rise to 85%, and an F1-score enhancement to 85.5%. These figures significantly surpass those achieved by models utilizing standard adversarial training (75% accuracy) and defensive distillation (70% accuracy), showcasing IAT’s superior capacity to maintain model accuracy under adversarial conditions.In conclusion, IAT stands out as an effective defense mechanism, significantly bolstering the resilience of CNN-LSTM models against adversarial perturbations. This research not only sheds light on the vulnerabilities of these models to adversarial attacks but also establishes IAT as a benchmark in defense strategy development, promising enhanced security and reliability for PQD classification and related applications.

Author 1: Ganesh Ingle

Author 2: Sanjesh Pawale

Keywords: Adversarial attacks; Input Adversarial Training (IAT); deep learning security; model robustness

Paper 121: Image Binary Matrix Processing to Encrypt-Decrypt Digital Images

Abstract: This research study presents a simple cryptographic solution for protecting grayscale and colored digital images, which are commonly used in computer applications. Due to their widespread use, protecting these photos is crucial to preventing unauthorized access. This article’s methodology manipulates an image’s binary matrix using basic operations. These specified actions include increasing the 8-column matrix to 64 columns, reorganizing it into 64 columns, separating it into four blocks, and shuffle the columns using secret index keys. These keys are produced using four sets of common chaotic logistic parameters. Each set executes a chaotic logistic map model to generate a chaotic key, which is then translated into an index key. This index key shuffles columns during encryption and reverses during decryption. The cryptographic approach promises a large key space that can withstand hacking. The encrypted image is secure since the decryption procedure is sensitive to the precise private key values. Private keys are frequently chaotic logistic parameters, making encryption resilient. This method is convenient since it supports images of any size and kind without modifying the encryption or decryption techniques. Shuffling replaces difficult logical procedures in typical data encryption methods, simplifying the cryptographic process. Experiments with several photos will evaluate the proposed strategy. The encrypted and decrypted photos will be examined to ensure the method meets cryptographic standards. Speed tests will also compare the proposed method to existing cryptographic methods to show its potential to speed up picture cryptography by lowering encryption and decryption times.

Author 1: Mohamad Al-Laham

Author 2: Firas Omar

Author 3: Ziad A. Alqadi

Keywords: Image processing; binary matrix; encrypt-decrypt; digital image

Paper 122: An End-to-End Model of ArVi-MoCoGAN and C3D with Attention Unit for Arbitrary-view Dynamic Gesture Recognition

Abstract: Human gesture recognition is an attractive research area in computer vision with many applications such as human-machine interaction, virtual reality, etc. Recent deep learning techniques have been efficiently applied for gesture recognition, but they require a large and diverse amount of training data. In fact, the available gesture datasets contain mostly static gestures and/or certain fixed viewpoints. Some contain dynamic gestures, but they are not diverse in poses and viewpoints. In this paper, we propose a novel end-to-end framework for dynamic gesture recognition from unknown viewpoints. It has two main components: (1) an efficient GAN-based architecture, named ArVi-MoCoGAN; (2) the gesture recognition component, which contains C3D backbones and an attention unit. ArVi-MoCoGAN aims at generating videos at multiple fixed viewpoints from a real dynamic gesture at an arbitrary viewpoint. It also returns the probability that a real arbitrary view gesture belongs to which of the fixed-viewpoint gestures. These outputs of ArVi-MoCoGAN will be processed in the next component to improve the arbitrary view recognition performance through multi-view synthetic gestures. The proposed system is extensively analyzed and evaluated on four standard dynamic gesture datasets. The experimental results of our proposed method are better than the current solutions, from 1% to 13.58% for arbitrary view gesture recognition and from 1.2% to 7.8% for single view gesture recognition.

Author 1: Huong-Giang Doan

Author 2: Hong-Quan Luong

Author 3: Thi Thanh Thuy Pham

Keywords: Dynamic gesture recognition; attention unit; generative adversarial network

Paper 123: Predicting ICU Admission for COVID-19 Patients in Saudi Arabia: A Comparative Study of AdaBoost and Bagging Methods

Abstract: COVID-19’s high fatality rate and accurately deter-mining the mortality rate within a particular geographic region continue to be significant concerns. In this study, the authors investigated and assessed the performance of two advanced machine learning approaches, Adaptive Boosting (AdaBoost) and Bootstrap Aggregation (Bagging), as strong predictors of COVID- 19-related intensive care unit (ICU) admissions within Saudi Arabia. These models may help Saudi health-care organizations determine who is at a higher risk of readmission, allowing for more targeted interventions and improved patient outcomes. The authors found AdaBoost-RF and Bagging-RF methods produced the most precise models, with accuracy rates of 97.4% and 97.2%, respectively. This work, like prior studies, illustrates the viability of developing, validating, and using machine learning (ML) prediction models to forecast ICU admission in COVID-19 cases. The ML models that have been developed have tremendous potential in the fight against COVID-19 in the health-care industry.

Author 1: Hamza Ghandorh

Author 2: Mohammad Zubair Khan

Author 3: Mehshan Ahmed Khan

Author 4: Yousef M. Alsofayan

Author 5: Ahmed A. Alahmari

Author 6: Anas A. Khan

Keywords: COVID-19; adaptive boosting; bootstrap aggregation; prediction; ICU admission; Saudi Arabia; machine learning

Paper 124: Secure Sharing of Patient Controlled e-Health Record using an Enhanced Access Control Model with Encryption Based on User Identity

Abstract: Healthcare industry is converting to digital due to the constantly evolving medical needs in the modern digital age. Many researchers have put up models like Ciphertext Policy Attribute Based Encryption (CPABE) to provide security to health records. But, the CPABE-variants failed to give total control of a medical record to its corresponding owner i.e., patient. Recently, Mittal et al. suggested that Identity Based Encryption (IBE) can be used to achieve this. But, this model used a Key Generation Center (KGC) to maintain keys that reduces the trust as the keys may get leaked. To overcome this problem, an enhanced access control model along with data encryption is presented where a separate key generation center is not needed. Because of this, the processing time for setting-up and extraction of keys is minimized. The total processed time of proposed is 74.42ms. But, the same is 92.89ms, 165.42ms, and 218.75ms in case of Boneh-Franklin, Zhang et al., and Yu et al., respectively. Our proposed model also gives a patient the complete control of his/her own health record. The data owner can decide who can access the record (full/ partial) with what access rights (read/write/ update). The data requestors can be a doctor/ nurse/insurance providers/ researchers and so on. The requestors are not based on groups or roles but based on an identity that is accepted by the data owner. The proposed model also withstands the key leakage attacks that are due to the key generation center.

Author 1: Mohinder Singh B

Author 2: Jaisankar N

Keywords: Access permissions; fine-grained access control; identity based encryption; key generation center; electronic health record

Paper 125: Unmasking Fake Social Network Accounts with Explainable Intelligence

Abstract: The recent global social network platforms have intertwined a web connecting people universally, encouraging unprecedented social interactions and information exchange. However, this digital connectivity has also spawned the growth of fake social media accounts used for mass spamming and targeted attacks on certain accounts or sites. In response, carefully con-structed artificial intelligence (AI) models have been used across numerous digital domains as a defense against these dishonest accounts. However, clear articulation and validation are required to integrate these AI models into security and commerce. This study navigates this crucial turning point by using Explainable AI’s SHAP technique to explain the results of an XGBoost model painstakingly trained on a pair of datasets collected from Instagram and Twitter. These outcomes are painstakingly inspected, assessed, and benchmarked against traditional feature selection techniques using SHAP. This analysis comes to a head in a demonstrative discourse demonstrating SHAP’s suitability as a reliable explainable AI (XAI) for this crucial goal.

Author 1: Eman Alnagi

Author 2: Ashraf Ahmad

Author 3: Qasem Abu Al-Haija

Author 4: Abdullah Aref

Keywords: Explainable Artificial Intelligence (XAI); Shapley Additive exPlanations (SHAP); feature selection; fake accounts detection; social media

Paper 126: The Effect of Pre-processing on a Convolutional Neural Network Model for Dorsal Hand Vein Recognition

Abstract: There are numerous techniques for identifying users, including cards, passwords, and biometrics. Emerging technologies such as cloud computing, smart gadgets, and home automation have raised users’ awareness of the privacy and security of their data. The current study aimed to utilise the CNN model augmented with various pre-processing filters to create a reliable identification system based on the DHV. In addition, the proposed implementing several pre-processing filters to enhance CNN recognition accuracy. The study used a dataset of 500 hand-vein images extracted from 50 patients, while the dataset training was done using the data augmentation technique. The accuracy of the proposed model in this study in classifying images without using image processing showed that 70% was approved for training. Moreover, the results indicated that using the mean filter to remove the noise gave better results, as the accuracy reached 99% in both training conditions.

Author 1: Omar Tarawneh

Author 2: Qotadeh Saber

Author 3: Ahmed Almaghthawi

Author 4: Hamza Abu Owida

Author 5: Abedalhakeem Issa

Author 6: Nawaf Alshdaifat

Author 7: Ghaith Jaradat

Author 8: Suhaila Abuowaida

Author 9: Mohammad Arabiat

Keywords: CNNs; preprocessing; dorsal hand vein; recognition; CNN; authentication

Paper 127: A Machine Learning-based Solution for Monitoring of Converters in Smart Grid Application

Abstract: The integration of renewable energy sources and the advancement of smart grid technologies have revolutionized the power distribution landscape. As the smart grid evolves, the monitoring and control of power converters play a crucial role in ensuring the stability and efficiency of the overall system. This research paper introduced a converter monitoring system in photovoltaic systems, the main concern is to protect the electrical system from disastrous failures that occur when the system is in operating condition. The reliability of the converters is significantly influenced by the degradation of their passive components, which can be characterized in various ways. For instance, the aging of inductors and capacitors can be char-acterized by a decrease in their inductance and capacitance values. Identifying which component is undergoing degradation and assessing whether it is in a critical condition or not, is crucial for implementing cost-effective maintenance strategies. This paper explores a set of classification algorithms, leveraging machine learning, trained on data collected from a Zeta converter simulated in Matlab Simulink. the report presents observations on how each algorithm effectively predicts the component and its condition and Graphical Performance Comparison for different ML Techniques serves as a crucial endeavor in evaluating and understanding the effectiveness of various ML approaches. The goal is to provide a comprehensive overview of how these techniques fare concerning criteria such as accuracy, precision, recall, F1 score, and Specificity among others. Quadratic Support Vector Machine (SVM) yields superior results compared to other machine learning techniques employed in training our dataset.

Author 1: Umaiz Sadiq

Author 2: Fatma Mallek

Author 3: Saif Ur Rehman

Author 4: Rao Muhammad Asif

Author 5: Ateeq Ur Rehman

Author 6: Habib Hamam

Keywords: Artificial intelligence; photovoltaic; support vector machine; machine learning; K-Nearest neighbor; maximum power point tracking; pulse width modulation; prognostic analysis; one-against-rest; one-against-one; direct acyclic graph; multi class support vector machine; DC-DC converter; zeta c

Paper 128: Robust Stability Analysis of Switched Neutral Delayed Systems with Parameter Uncertainties

Abstract: A time-delay neural system is an accurate class of neural system that exposes delays in both the state values and their derivatives. In this case, it is critical to maintain the system stability. Here, the stability investigation on uncertain switched-neutral systems with state-time delays is the focus of this paper. In fact, a novel adequate condition in terms of the feasibility of Linear Matrix Inequalities (LMIs) is offered to guarantee the global asymptotically stability of this category of systems with parameter uncertainties, based on the Lyapunov-Krasovskii functional method. Additionally, resistance against errors and disturbances can be ensured using the Multiple Quadratic Lyapunov Functions (MQLFs). Through a numerical example, the designed method’s effectiveness is proven.

Author 1: Nidhal Khorchani

Author 2: Rafika El Harabi

Author 3: Wiem Jebri Jemai

Author 4: Hassen Dahman

Keywords: Switched neutral systems; parameter uncertainties; delay-dependent; robust stability; multiple quadratic Lyapunov-Krasovskii; LMI technique

Paper 129: Video-based Domain Generalization for Abnormal Event and Behavior Detection

Abstract: Surveillance cameras have been widely deployed in public and private areas in recent years to enhance security and ensure public safety, necessitating the monitoring of unforeseen incidents and behaviors. An intelligent automated system is essential for detecting anomalies in video scenes to save the time and cost associated with manual detection by laborers monitoring displays. This study introduces a deep learning method to identify abnormal events and behaviors in surveillance footage of crowded areas, utilizing a scene-based domain generalization strategy. By utilizing the keyframe selection approach, keyframes containing relevant information are extracted from video frames. The chosen keyframes are utilized to create a spatio-temporal entropy template that reflects the motion area. The acquired template is then fed into the pre-trained AlexNet network to extract high-level features. The study utilizes the Relieff feature selection approach to choose suitable features, which are then served as input to Support Vector Machine (SVM) classifier. The model is assessed using six available datasets and two datasets built in this research, containing videos of normal and abnormal events and behaviors. The study found that the proposed method, utilizing domain generalization, surpassed state-of-arts methods in terms of detection accuracy, achieving a range from 87.5% to 100%. It also demonstrated the model’s effectiveness in detecting anomalies from various domains with an accuracy rate of 97.13%.

Author 1: Salma Kammoun Jarraya

Author 2: Alaa Atallah Almazroey

Keywords: Domain generalization; abnormal event; abnormal behavior

Paper 130: An Efficient Blockchain Neighbor Selection Framework Based on Agglomerative Clustering

Abstract: Blockchain-based decentralized applications have garnered significant attention and have been widely deployed in recent years. However, blockchain technology faces several challenges, such as limited transaction throughput, large blockchain sizes, scalability, and consensus protocol limitations. This paper introduces an efficient framework to accelerate broadcast efficiency and enhance the blockchain system’s throughput by reducing block propagation time. It addresses these concerns by proposing a dynamic and optimized Blockchain Neighbor Selection Framework (BNSF) based on agglomerative clustering. The main idea behind the BNSF is to divide the network into clusters and select a leader node for each cluster. Each leader node resolves the Minimum Spanning Tree (MST) problem for its cluster in parallel. Once these individual MSTs are connected, they form a comprehensive MST for the entire network, where nodes obtain optimal neighbors to facilitate the process of block propagation. The evaluation of BNSF showed superior performance compared to neighbor selection solutions such as Dynamic Optimized Neighbor Selection Algorithm (DONS), Ran-dom Neighbor Selection (RNS), and Neighbor Selection based on Round Trip Time (RTT-NS). Furthermore, BNSF significantly reduced the block propagation time, surpassing DONS, RTT-NS, and RNS by 51.14%, 99.16%, and 99.95%, respectively. The BNSF framework also achieved an average MST calculation time of 27.92% lower than the DONS algorithm.

Author 1: Marwa F. Mohamed

Author 2: Mostafa Elkhouly

Author 3: Safa Abd El-Aziz

Author 4: Mohamed Tahoun

Keywords: Blockchain; scalability; agglomerative clustering; broadcasting; optimized neighbor selection; minimum spanning tree; parallel processing

Paper 131: Unified Access Management for Digital Evidence Storage: Integrating Attribute-based and Role-based Access Control with XACML

Abstract: Digital evidence is stored in digital evidence storage. An access control system is crucial in situations where not all users can access digital evidence, ensuring that each user’s access is limited to what is essential for them to do their jobs. As a result, access control must be included. Role-based access control (RBAC) and attribute-based access control (ABAC) are two of the several varieties of access control. Only the ABAC model is applied in digital evidence storage systems in the research that has been done. In order to get more precise findings, some academics have suggested combining these two models. In light of this, this study suggests a hybrid paradigm for digital evidence storage that combines the key components of both ABAC and RBAC. In addition to utilizing eXtensible Access Control Markup (XACML) throughout the policy statement creation process. A programming language called XACML uses the XML format to specify RBAC and ABAC rules. The study’s findings demonstrate that the ABAC and RBAC models can function in accordance with the developed permit and deny test scenarios.

Author 1: Ayu Maulina

Author 2: Zulfany Erlisa Rasjid

Keywords: ABAC; RBAC; digital evidence storage; XACML; network security

Paper 132: Detecting and Visualizing Implementation Feature Interactions in Extracted Core Assets of Software Product Line

Abstract: Recently, software products have played a vital role in our daily lives, having a significant impact on industries and the economy. Software product line engineering is an engineering strategy that allows for the systematic reuse and development of a set of software products simultaneously, rather than just one software product at a time. This strategy mainly relies on features composition to generate multiple new software products. Unwanted feature interactions, where the integration of multiple feature implementations hinders each other, are challenging in this strategy. This leads to performance degradation, and unexpected behaviors may happen. In this article, we propose an approach to detect and visualize all feature interactions early. Our approach depends on an unsupervised clustering technique called formal concept analysis to achieve the goal. The effectiveness of the proposed approach is evaluated by applying it to a large and benchmark case study in this domain. The results indicate that the proposed approach effectively detects and visualizes all interacted features. Also, it saves developer efforts for detecting interacted features in a range between 67% and 93%.

Author 1: Hamzeh Eyal Salman

Author 2: Yaqin Al-Ma’aitah

Author 3: Abdelhak-Djamel Seriai

Keywords: Unwanted feature interaction; core assets; extractive approach; visualization; shared artifacts; implementation dependency

Paper 133: An Optimized Air Traffic Departure Sequence According to the Standard Instrument Departures

Abstract: Sequencing efficiently the departure traffic remains among the critical parts of air traffic management these days. It not only reduces delays and congestion at hold points, but it also enhances airport operations, improves traffic planning, and increases capacity. This research paper proposes an approach, that employs a genetic algorithm (GA), to help air traffic con-trollers in organizing a sequence for the departure traffic based on the standard instrument departures (SIDs) configuration. A scenario with Randomly assigned types, SIDs, and departure times was applied to a set of aircraft in a terminal area with a four-SID configuration to assess the performance of the suggested GA. Subsequently, a comparison with the standard method of First Come First Served (FCFS) was conducted. The testing data revealed promising results in terms of the total spent time to reach a specified altitude after takeoff.

Author 1: Abdelmounaime Bikir

Author 2: Otmane Idrissi

Author 3: Khalifa Mansouri

Author 4: Mohamed Qbadou

Keywords: Air traffic management; standard instrument departure; departure traffic sequencing; genetic algorithm; heuristic algorithm

Paper 134: Utilizing Various Machine Learning Techniques for Diabetes Mellitus Feature Selection and Classification

Abstract: Diabetes mellitus is a chronic disease affecting over 38.4 million adults worldwide. Unfortunately, 8.7 million were undiagnosed. Early detection and diagnosis of diabetes can save millions of people’s lives. Significant benefits can be achieved if we have the means and tools for the early diagnosis and treatment of diabetes since it can reduce the ratio of cardiovascular disease and mortality rate. It is urgently necessary to explore computational methods and machine learning for possible assistance in the diagnosis of diabetes to support physician decisions. This research utilizes machine learning to diagnose diabetes based on several selected features collected from patients. This research provides a complete process for data handling and pre-processing, feature selection, model development, and evaluation. Among the models tested, our results reveal that Random Forest performs best in accuracy (i.e., 0.945%). This emphasizes Random Forest’s efficiency in precisely helping diagnose and reduce the risk of diabetes.

Author 1: Alaa Sheta

Author 2: Walaa H. Elashmawi

Author 3: Ahmad Al-Qerem

Author 4: Emad S. Othman

Keywords: Diabetes; machine learning; random forest; SMOTE technique

Paper 135: Screening Cyberattacks and Fraud via Heterogeneous Layering

Abstract: On the Internet of Things (IoT) age, intelligent equipment is employed to give effective and dependable utilization of applications. IoT devices may recognize and provide extensive information while also intelligently processing that data. Data systems, systems for control, plus sensing are growing increasingly vital in contemporary manufacturing processes. The amount of internet of things gadgets and methods used is growing, that has culminated in a rise in assaults. Such assaults have the potential to interrupt international activities and cause major financial losses. Multiple methods, including Machine learning (ML) in addition to Deep Learning (DL), are being utilized for identifying cyberattack. In this investigation, researchers offer an ensemble staking approach that is strong strategy in ML for detecting assaults via the Internet of Things having excellent accuracy. Tests were carried out using three distinct information: credit card data, NSL-KDD, and UNSW. Single fundamental classifications were beaten by the suggested layered ensembles classification. The results show that the cyberattack detection model in this research possessed a 95.15% accuracy percentage, while the credit card fraud detection model achieved a 93.50% accuracy percentage.

Author 1: Abdulrahman Alahmadi

Keywords: Fraud; Internet of Things (IoT); Deep Learning (DL); ensemble; stacking; cyberattack; Machine Learning (ML)

Paper 136: DeepSL: Deep Neural Network-based Similarity Learning

Abstract: The quest for a top-rated similarity metric is inherently mission-specific, with no universally ”great” metric relevant across all domain names. Notably, the efficacy of a similarity metric is regularly contingent on the character of the challenge and the characteristics of the records at hand. This paper introduces an innovative mathematical model called MCESTA, a versatile and effective technique designed to enhance similarity learning via the combination of multiple similarity functions. Each characteristic within it is assigned a selected weight, tailor-made to the necessities of the given project and data type. This adaptive weighting mechanism enables it to outperform conventional methods by providing an extra nuanced approach to measuring similarity. The technique demonstrates significant enhancements in numerous machine learning tasks, highlighting the adaptability and effectiveness of our model in diverse applications.

Author 1: Mohamedou Cheikh Tourad

Author 2: Abdali Abdelmounaim

Author 3: Mohamed Dhleima

Author 4: Cheikh Abdelkader Ahmed Telmoud

Author 5: Mohamed Lachgar

Keywords: Similarity learning; Siamese networks; MCESTA; triplet loss; similarity metrics

Paper 137: Unveiling the Dynamic Landscape of Malware Sandboxing: A Comprehensive Review

Abstract: In contemporary times, the landscape of malware analysis has advanced into an era of sophisticated threat detection. Today’s malware sandboxes conduct rudimentary analyses and have evolved to incorporate cutting-edge artificial intelligence and machine learning capabilities. These advancements empower them to discern subtle anomalies and recognize emerging threats with a heightened level of accuracy. Moreover, malware sandboxes have adeptly adapted to counteract evasion tactics, creating a more realistic and challenging environment for malicious entities attempting to detect and evade analysis. This paper delves into the maturation of malware sandbox technology, tracing its progression from basic analysis to the intricate realm of advanced threat hunting. At the core of this evolution is the instrumental role played by malware sandboxes in providing a secure and dynamic environment for the in-depth examination of malicious code, contributing significantly to the ongoing battle against evolving cyber threats. In addressing the ongoing challenges of evasive malware detection, the focus lies on advancing detection mechanisms, leveraging machine learning models, and evolving malware sandboxes to create adaptive environments. Future efforts should prioritize the creation of comprehensive datasets, distinguish between legitimate and malicious evasion techniques, enhance detection of unknown tactics, optimize execution environments, and enable adaptability to zero-day malware through efficient learning mechanisms, thereby fortifying cybersecurity defences against emerging threats.

Author 1: Elhaam Debas

Author 2: Norah Alhumam

Author 3: Khaled Riad

Keywords: Malware analysis; threat hunting; security operations; machine learning; cutting-edge AI; sandboxing

Paper 138: Enhancing Low-Resource Question-Answering Performance Through Word Seeding and Customized Refinement

Abstract: The state-of-the-art approaches in Question-Answering (QA) systems necessitate extensive supervised training datasets. In low-resource languages (LRL), the scarcity of data poses a bottleneck, and the manual annotation of labeled data is a rigorous process. Addressing this challenge, some recent efforts have explored cross-lingual or multilingual QA learning by leveraging training data from resource-rich languages (RRL). However, the efficiency of such approaches relies on syntactic compatibility between languages. The paper introduces the innovative method that involves seeding LRL data into RRL to create a bilingual supervised corpus while preserving the syntactical structure of RRL. The method employs the translation and transliteration of selected parts-of-speech (POS) category words. Additionally, the paper also proposes a customized approach to fine-tune the models using bilingual data. Employing the bilingual data and the proposed fine-tuning approach, the most successful model has achieved a 75.62 F1 score on the XQuAD Hindi dataset and a 68.92 F1 score on the MLQA Hindi dataset in a zero-shot architecture. In the experiments conducted using few-shot learning setup, the highest F1 scores of 79.17 on the XQuAD Hindi dataset and 70.42 on the MLQA Hindi dataset have been achieved.

Author 1: Hariom Pandya

Author 2: Brijesh Bhatt

Keywords: Embedding learning; words seeding; bilingual dataset generation; low-resource question-answering