		FeedBack \| Contact Us \| Links \| Site Map

Welcome to Journal of Theoretical and Applied Information Technology

Submit Paper / Call for Papers

Journal receives papers in continuous flow and we will consider articles from a wide range of Information Technology disciplines encompassing the most basic research to the most innovative technologies. Please submit your papers electronically to our submission system at http://jatit.org/submit_paper.php in an MSWord, Pdf or compatible format so that they may be evaluated for publication in the upcoming issue. This journal uses a blinded review process; please remember to include all your personal identifiable information in the manuscript before submitting it for review, we will edit the necessary information at our side. Submissions to JATIT should be full research / review papers (properly indicated below main title).

Journal of Theoretical and Applied Information Technology

August 2024 | Vol. 102 No.15

Title:	OPTIMIZING BROKERAGE COMPANY MARKETING: A WEB-BASED PERFORMANCE MONITORING SYSTEM WITH CODEIGNITER
Author:	ANGGA ADITYA PERMANA , FEBI LATIFAH
Abstract:	PT Kontrak Perkasa Futures provide trading services, particularly commodities that generate various investment products. The company's operations rely on the performance of the marketing team, especially in the process of acquiring customers willing to have their investment funds managed. Currently, the performance reporting process for the marketing team employees at PT. Kontak Perkasa Futures is still carried out conventionally through manual recording in a ledger. This is very ineffective and inefficient for the employee especially in real time reporting process to their manager. This research aims to design a web-based information system using the CodeIgniter Framework to assist in computerizing the reporting process so that the performance of the marketing team can be monitored. The research methodology involves requirement analysis, system design, system development and system evaluation. The data collection methodology employed in this research includes interviews and observations, with the data analyzed using performance, information, economy, control, efficiency, and service (PIECES) analysis. The design of the marketing performance information system utilizes unified modeling language (UML). The system development method is using Scrum framework and built using MySQL, CodeIgniter Framework and PHP language. The system evaluation is using black box testing. The result of this research is a recommendation for the design of a marketing performance information system that can be used by the company as a reference for the development of marketing performance information systems in the future. Besides that, designing a web-based information system can make positive contribution in the effectiveness and efficiency of data and information management.
Keywords:	Black Box Testing. Codeigniter, Performance, SCRUM, UML.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	REPLICATING VIDEO GAME PLAYERS' BEHAVIOR THROUGH DEEP REINFORCEMENT LEARNING ALGORITHMS
Author:	HAFSA GHARBI, MR. LOTFI ELAACHAK, AND MR. ABDELHADI FENNAN
Abstract:	This paper addresses the challenge of imitating the behavior of video game players using Deep Reinforcement Learning algorithms. By training intelligent agents with algorithms such as Proximal Policy Optimization (PPO), Behavioral Cloning (BC), and Generative Adversarial Imitation Learning (GAIL), we enable these agents to learn from their interactions with the game environment, optimizing their actions based on rewards and punishments. Experimental evaluations across various video games demonstrate that these trained agents can successfully mimic human player behavior in complex situations. This capability offers significant opportunities for creating challenging non-player characters (NPCs), designing adaptive difficulty levels, and enhancing the overall gaming experience. Our findings suggest that integrating Reinforcement Learning techniques allows game developers to provide more realistic and immersive gameplay, effectively bridging the gap between Artificial Intelligence and both video games and serious games.
Keywords:	Deep Reinforcement Learning, Proximal Policy Optimization, Behavioral cloning, Generative Adversarial Imitation Learning, video game.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	COMBATING FRAUD: DYNAMIC AND ADVANCED TECHNIQUES FOR UNVEILING FALSE REVIEWS AND DECEIVING TEXT ON E-COMMERCE WEBSITE
Author:	DR. P. NAGARAJ, RAJASRI MAMIDALA
Abstract:	E-commerce has widely grown among people in recent years and has been used for purchasing products and services on the Internet. E-commerce faces more challenges due to the growing amount of deceptive and fake products online. This research aims to combat this fraud using dynamic and advanced techniques for unveiling false reviews and deceiving product descriptions. This research employs the DistilBERT model for detecting fake reviews and the BERT base model for identifying misleading product descriptions. This research aims to bring down false information, stop defying products backed up by their false reviews, and report them. In this study, we create an FRD Algorithm and a DTD Algorithm that solve the problem of Combating fraud: dynamic and advanced Techniques for unveiling false reviews and deceiving text on e-commerce websites. The model is achieving an accuracy of 95.6 %. It helps customers save more time and focus on purchasing the product rather than determining whether the reviews are true or false. Future research focuses on a more accurate, dynamic, and efficient way to execute the AI models.
Keywords:	Fake Review Detection, Deceiving Text Detection, BERT Model, FRD Algorithm, DTD Algorithm.
Source:	Journal of Theoretical and Applied Informaton Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	IMPLEMENTATION AND MANAGEMENT OF SECURITY FOR SENSITIVE DATA IN CLOUD COMPUTING ENVIRONMENT USING ELLIPTICAL CURVE CRYPTOGRAPHY
Author:	N. KRISHNAMOORTHY, S.UMARANI
Abstract:	When it comes to cloud storage, one of the biggest concerns is keeping data safe. In today's technologically evolved world, cloud attacks are on the rise. Current cloud storage security services rely heavily on symmetric key encryption algorithms, which involve the exchange of secret keys and may be susceptible to attacks from outside parties. In the same way that cloud computing has become indispensable in recent years, so too have the security issues surrounding the cloud model grown in scope and complexity. There has been a dramatic shift in how infrastructure, service delivery, and development models are seen thanks to cloud computing. Each participant in the proposed cloud storage security framework has a specific role to play: the data owner, who is responsible for encrypting the plain text and constructing the access control policy; the attribute authority, who acts as the data owner's trusted agent and stores the attribute-based access control policy used for key generation and to restrict access to authorized users; the cloud storage; and the data users. In Elliptic Curve Cryptography (ECC), improving encoding schemes is a top priority. Despite the obvious advantages of the cloud model, it will have a hard time gaining widespread client acceptance unless issues of privacy and security are resolved. Concerns about the privacy, authenticity, and integrity of cloud-stored data are the focus of this thesis, along with recommendations for implementing those safeguards. In this thesis, we design and develop secure and efficient protocols based on Elliptic Curve Cryptography for multilevel security in a cloud environment, with the goals of protecting the privacy of users' data, ensuring that only authorized users can access their information, and guaranteeing that all data is genuine and unaltered. When it comes to encryption, computational load, and security during cloud data storage, retrieval, and access, the proposed security frameworks have been implemented and compared to existing models. This research work aims at introducing four different novel algorithms and finds out the minimized encoding and decoding durations and thereby reducing uploading and downloading durations.
Keywords:	Elliptic Curve Cryptography, Data Security, Cloud deployment, Private cloud, Public cloud
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	BIG DATA AND ARTIFICIAL INTELLIGENCE IN HIGHER EDUCATION: IMPACTS ON RESEARCH INTO NUTRITION OF CANCER PATIENTS
Author:	DARIFA EL HAIRACH, SOAD KHAL-LAYOUN, ABDELLATIF BOUR
Abstract:	The integration of Big Data and Artificial Intelligence (AI) technologies has brought significant transformations to higher education, impacting research, university management, and student experience. This paper explores the multifaceted influence of Big Data and AI within the higher education landscape. Beginning with an introduction, the paper delineates the growing importance of these technologies. It then delves into the role of Big Data, elucidating its definition, significance, and applications in both research and university management. Likewise, the section on Artificial Intelligence discusses its definition, importance, and its role in enhancing student learning and administrative processes. Moreover, the paper explores the synergy between Big Data and AI, discussing the benefits, challenges, and ethical considerations inherent in their integration. Highlighting successful integration examples, it underscores the transformative potential of combining these technologies in higher education contexts. Moving forward, the paper examines the impacts of Big Data and AI on research, emphasizing their role in accelerating data analysis, enhancing predictive modeling, and facilitating collaboration among researchers. Furthermore, the paper presents the results of the integration of these technologies on university management and student experience. It discusses how Big Data and AI have streamlined administrative processes and personalized student learning experiences. Through a comparative analysis with a case study, the paper offers insights into the real-world application and effectiveness of these technologies in higher education settings. The paper underscores the profound impacts of Big Data and AI on higher education, advocating for their continued integration to foster innovation, efficiency, and personalized learning experiences for students and researchers alike. In addition to the broader impacts on higher education, this paper includes a focused case study on the application of Big Data and AI in nutrition research for cancer patients. This case study demonstrates the transformative potential of these technologies in a specialized field of study. By leveraging Big Data, researchers can analyze vast datasets from clinical trials, patient records, and nutritional studies to identify patterns and correlations that were previously undetectable. AI algorithms can then be used to develop predictive models that help in personalizing nutritional plans for cancer patients, optimizing their treatment outcomes.
Keywords:	Big Data, Artificial Intelligence, Higher Education, Research, patient Experience
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	ENHANCING VISUAL CONTENT QUALITY IN EDUCATION AND RELIGION: A FUSION APPROACH OF MIRNET-V2 AND AUTOENCODERS FOR IMAGE PREPROCESSING AND NOISE REDUCTION
Author:	SMITA RATH1 , SUSHREE BIBHUPRADA B. PRIYADARSHINI, PRABHAT KUMAR SAHU, MONALISA PANDA, DEEPAK KUMAR PATEL, NIBEDITA JAGADEV , SIPRA SAHOO, NARAYAN PATRA
Abstract:	Combining MIRNet-v2, an advanced image preprocessing technique, with autoencoders for noise reduction can have implications beyond traditional image processing applications. This study could be related to education and religion journals. High-quality images are essential for conveying complex information effectively. Research on advanced image processing techniques like MIRNet-v2 combined with autoencoders could focus on improving the quality of educational images used in textbooks, online courses, or presentations. Therefore, a study combining MIRNet-v2 and autoencoders for image preprocessing and noise reduction could find relevance in journals focusing on education technology, digital humanities, religious studies, or interdisciplinary research at the intersection of technology and society. MIRNetv2 can help print illustrations and book photos by color correction, which involves adjusting colors to improve their accuracy and vibrancy. It will enhance contrast to improve the visual attractiveness of photographs, and low-light enhancement will add brightness to very dark photos. Machine learning algorithms have shown promising results in noise reduction tasks by learning the statistical characteristics of the noise and the underlying image structures. The central component of our method is a multi-scale residual block that includes several crucial components: (a) parallel multi-resolution convolution flows for gathering multi-scale characteristics; (b) exchange of data across the multi-resolution streams; (c) spatial and channel attention processes for preserving contextual data; and (d) attention-based multi-scale feature accumulation. The proposed approach involves collecting relevant image datasets, preprocessing them, selecting suitable machine learning algorithms, optimizing the model parameters, validating and evaluating the models, and identifying potential areas for future research and improvements. Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), and Mean Squared Error (MSE) are the three famous performance metrics frequently used in image processing.In contrast to PSNR, which emphasizes pixel-level variations, SSIM considers the image data's structure and brightness. The luminance, contrast, and structure of the processed and original images are compared. The results clearly show that MIRNet-v2 outperforms previous approaches by a significant margin. MIRNet-v2 achieves a total performance gain of 3.44 dB on the LoL, SIDD, and DND datasets and some real-time images as training and testing datasets. In summary, this method trains an extended set of features that combines contextual data from many scales while maintaining high-resolution spatial aspects. Extensive tests on live picture benchmark datasets show that MIRNet-v2 produces cutting-edge performance for various image processing applications, comprising super-resolution, image denoising, and image enhancement.
Keywords:	Auto encoders; Convolutional Neural Network; MIRNet-v2; Noise Reduction; Deep Learning
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	DIGITAL SECURITY IN MOROCCAN UNIVERSITIES IN THE ERA OF ARTIFICIAL INTELLIGENCE
Author:	M.BOUJARRA, A.AL KARKOURI, 1 Y.FAKHRI, S.BOUREKKADI
Abstract:	The rapid advancement of information technology has significantly influenced the digital security environment in Moroccan universities, compelling these institutions to reconsider their approaches in response to more complex risks. The growing digitalization of academic data, including student records and research projects, underscores the crucial need of safeguarding these digital assets, which have become indispensable foundations of contemporary academic endeavors. Artificial intelligence (AI) integration is becoming a critical solution to address the increasing security problems. AI provides a range of sophisticated features, such as proactive identification of threats, immediate analysis of behavior, and the capacity to preemptively stop possible assaults before they occur. Nevertheless, the implementation of heightened security measures presents both practical and ethical obstacles. From a logistical standpoint, the successful implementation of AI necessitates the presence of suitable technical infrastructure and a workforce that is adequately educated to fully use this technology. The expenses linked to the implementation of AI-driven solutions, while potentially lucrative in the future, are a significant concern for Moroccan institutions with sometimes constrained financial resources. Simultaneously, the need to evaluate and control ethical hazards linked to the utilization of AI, such as algorithmic prejudice and privacy ramifications, introduces an intricate aspect to this technological shift. at order to provide a background for these difficulties, the essay explores the present state of digital security at Moroccan institutions, highlighting particular risks and weaknesses that already exist. Subsequently, it examines practical implementations of AI in the academic sphere, showcasing how this technology might be used preemptively to enhance safeguards against cyber hazards. The paper emphasizes the need of a strategy tailored to the local environment by addressing the issues unique to Moroccan institutions, such as financial limitations and training requirements. The suggested answers are derived on a comprehensive comprehension of the distinct requirements of each institution, emphasizing practical approaches to surmount these challenges. Ultimately, a pragmatic case study demonstrates the triumphant integration of AI in a Moroccan university, providing tangible instances and valuable insights gained. These pragmatic observations seek to provide guidance to other Moroccan educational institutions in their pursuit of improved digital security, emphasizing the significance of adjustment, prudent allocation of resources, and the incorporation of ethical issues at every stage of the process.
Keywords:	Digital Security, Artificial Intelligence, Moroccan Universities, Cyber Threats, Technological Solutions
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	PREDICTION MODEL APPLYING MACHINE LEARNING TO FORECAST THE BANKRUPTCY OF COMPANIES. A SYSTEMATIC REVIEW OF THE LITERATURE
Author:	DINO QUINTEROS-NAVARRO , CIRO RODRIGUEZ
Abstract:	Prediction models that are aimed at companies allow trends to be identified and generate a much broader picture, making decisions more effective and efficient. In that sense, the research study uses a literature review to identify the state of the art of how predictive models can interact and generate more accurate and reliable results to identify, forecast and eventually reduce the bankruptcy of companies. In addition, the systematic analysis of the contributions was carried out where the main models of supervised learning and other techniques were considered. The authors [25] specify that a systematic literature review is a primary tool for developing an evidence base by identifying, evaluating, and interpreting all available research relevant to a particular research question, thematic area, or phenomenon of interest. Likewise, questions were posed through three (3) stages: identification of parameters, calculation of the ACCP and determination of factors. For the calculation of the weighted precision metric (ACCP), the most significant values of each model were used and the results obtained were analyzed. In this sense, the review must contain the following: method of analysis, theoretical basis, classification of payment yield prediction models and analysis of the topics according to the questions asked.
Keywords:	Bankruptcy, Predictive, Models, Machine Learning, Making Decisions
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	EARLY DIAGNOSIS OF GLAUCOMA BY OPTIC DISC AND OPTIC CUP SEGMENTATION OVER RETINAL FUNDUS IMAGES USING DEEP LEARNING ALGORITHM
Author:	R. GEETHALAKSHMI, R. VANI
Abstract:	Glaucoma is characterized by a progressive retinal ganglion cell loss and alters the optic nerve head in neuro retinal rim tissue, as well as restriction of the visual field. Glaucoma is a type of eye disorder that is the leading cause of permanent vision loss globally. The main symptom of glaucoma is the changes in the optical nerve head (ONH) which assist in the early recognition of glaucoma This misalignment of ONH reflects on Optic Disc (OD) and Optic Cup (OC) size and shape, creating changes in Cup to Disc Ratio (CDR). By calculating CDR, glaucoma can be diagnosed at early stage. In order to calculate CDR correctly, proper segmentation of the OD and OC is essential. In this proposed work an Adaptive median filter with TOPHAT is utilized for improving quality of the retinal image, further segmentation of OD and OC is performed by utilizing Multilevel Statistical Region of Interest (MSROI) and classification of Glaucoma is done by using a novel Artefact Convolutional Neural Network (ACNN). The accuracy of 99.38% is obtained using ACNN architecture. Further, dice coefficient of 0.79, Intersection Over Union (IOU) of 0.67, Mean of 73, Standard Deviation of 55 and Mean Square Error of 410 is achieved using MSROI based segmentation. The proposed work can be utilized as desktop application in early diagnosis of Glaucoma. This Computer Aided Diagnosis (CAD) method highly helpful in improving accuracy, reduce time consumption and also provide ease of diagnosing the disease since glaucoma require periodic diagnosis. The proposed model is trained and validated using Drion-DB dataset, the system provides robust and accurate result compared with the existing method.
Keywords:	Glaucoma, Segmentation, Optic Disc, Optic Cup, Artefact CNN, Deep Learning.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	DETECTING ONLINE GAME ADDICTION USING FUZZY LOGIC WITH THE TSUKAMOTO METHOD
Author:	MARYANA, NURDIN, BUSTAMI, FAJRIANA, 5NABILA ZULFANI, ARNAWAN HASIBUAN
Abstract:	Currently online games are not a stranger to people's lives in various circles, from children, adolescents and adults. The large number of people who are very fond of playing online games, the chance that people will become addicted to playing online games is very large, sufferers of online game addiction will not know the time so they often forget their obligations, in this case teenagers who are still in school become lazy and often skip school In order to become true gamers, so that bad things don't happen, a system model is needed to detect online game addiction. With the existence of this system model, it is hoped that it can provide information and solutions for people who have experienced addiction so that prevention and recovery can be carried out so that it does not have fatal consequences. This system model uses five input variables, namely thinking about games all day and feeling bad when not playing games, increased gaming time and loss of sleep, playing games to forget real life and ignore other activities, other people fail to prohibit playing games and will keep playing games, get annoyed easily and fight with family or friends, while the output variables consist of not addicted, mild addicted, heavy addicted. This study used 86 questionnaires filled in by 86 online game players. Then the results of the research were obtained, namely the level of not being addicted = 65%, mild addiction = 15% and heavy addiction = 20%.
Keywords:	Online Games, Input and Output Variables, Fuzzy Logic, Tsukamoto Method, Game Addiction Symptoms.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	DPARF: A DEEP LEARNING BASED INTELLIGENT FRAMEWORK FOR AUTOMATIC RECOGNITION OF PERSONAL ACTIVITY USING SMARTPHONE DATASET
Author:	BHAGYA REKHA SANGISETTI, SURESH PABBOJU
Abstract:	In the contemporary era, technological innovations such as Internet of Things (IoT) and Artificial Intelligence (AI) can offer unprecedented solutions to real world problems. IoT technology has paved way for sensor enabled data collection while AI enables learning and prediction to solve variety of problems. Personal Activity Recognition (PAR) is an important problem in many applications such as surveillance, security, computer gaming, sports, remote monitoring of humans or patients, healthcare, and military to mention few. Deep Learning (DL) techniques are widely used for solving PAR problem by processing data captured by sensors. IoT use cases or smart phones are capable of capturing accelerometer data that can be used to establish movements and thus recognize human activities. Existing methods based on deep learning witnessed success in PAR activities. Long Short-Term Memory (LSTM) is one such technique which could detect human activities with high accuracy. However, it is desired to improve it further towards leveraging prediction performance. Towards this end, in this paper, we proposed a framework known as Deep Personal Action Recognition Framework (DPARF) for automatic recognition of human activities. The framework is realized with our enhanced LSTM model known as CNN-5LSTM which is designed to improve accuracy in activity recognition. We proposed an algorithm known Enhanced LSTM with CNN for Automatic Personal Activity Recognition (ELSTM-CNN-APAR). This algorithm takes care of feature selection and processing of data consisting of temporal sequences. A standard PAR smartphone dataset from UCI repository is used in the empirical study. The experimental results revealed that our proposed model outperforms many existing models with 93.04% accuracy. It can be used to have an automated PAR Decision Support System (PAR-DSS) which may be integrated with a real time PAR system in question.
Keywords:	Personal Activity Recognition, Artificial Intelligence, Deep Learning, Enhanced LSTM, Smartphone Dataset
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	A DEEP LEARNING APPROACH TO SOFTWARE VULNERABILITY DETECTION BY LEVERAGING CNN’s AND COMPARING WITH RNN’s FOR IMPROVED ACCURACY AND EFFICIENCY
Author:	RAGHUPATHY DURGA PRASAD, Dr. MUKTEVI SRIVENKATESH
Abstract:	This study introduces an innovative paradigm for software security that leverages convolutional neural networks (CNNs) to detect emerging cyber threats. Our approach enhances software security detection mechanisms, offering superior performance compared to traditional machine learning methods and recurrent neural networks (RNNs), which underperformed in this context. The CNN architecture includes multiple convolutional layers for feature extraction, pooling layers for dimensionality reduction, and fully connected layers for classification, with non-linear activation functions like SoftMax to expedite classification. Dropout layers mitigate overfitting and enhance generalization. Using both synthetic and real-world data, our CNN model exhibited robust performance, achieving an accuracy of 0.91, precision of 0.90, recall of 0.89, and an F1-score of 0.895. These metrics indicate CNNs' proficiency in identifying intricate patterns and anomalies in software code, reducing false positives significantly. Although RNNs with LSTM or GRU layers capture temporal correlations in code sequences, they were less effective than CNNs in this application. The study's methodologies and code are available on Google Drive for cybersecurity specialists to replicate and build upon. By automating vulnerability detection with CNNs, cybersecurity professionals can focus more on pre-emptive measures. This research underscores the potential of CNNs to enhance software vulnerability detection, advocating for their integration with RNNs to create safer and more resilient software systems in response to escalating cyber threats.
Keywords:	Convolutional Neural Networks, Recurrent Neural Networks, Software Vulnerabilities, Automated Vulnerability Detection.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	HYBRID HENRY GAS-HARRIS HAWKS MODIFIED-OPPOSITION ALGORITHM FOR TASK SCHEDULING IN CLOUD COMPUTING
Author:	NORA OMRAN ALKAAM, ABU BAKAR MD. SULTAN, MASNIDA HUSSIN, AND KHAIRONI YATIM SHARIF
Abstract:	Objectives: Cloud computing environments allow users to remotely access computational resources and data computing services online. Task scheduling requires the development of reliable and efficient methods for mapping tasks to resources, making it an essential component of cloud computing. Effective task scheduling is critical for increasing operational efficiency since it entails carefully assigning tasks to resources to ensure optimal performance. This accurate coordination not only increases productivity, but it also optimizes resource allocation. Cloud computing solutions can improve overall system performance, reduce processing times, and increase efficiency by enhancing job scheduling. Methods: This study introduced a (Henry Gas-Harris Hawks Modified-Opposition) (HGHHM) algorithm to enhance the Henry Gas Solubility algorithm based on two components: Harris Hawks Optimization (HHO) and a modified comprehensive opposition-based learning (MOBL). This proposed HGHHM algorithm utilized the HHO method as a local search strategy to enhance the quality of approved solutions. While, MOBL improves the less effective solutions by carefully calculating their opposite equivalents and wisely choosing the most advantageous option. This approach facilitated the enhancement of suboptimal solutions, leading to an overall enhancement in the efficiency of the selected techniques. Results: CloudSim was used to test the HGHHM algorithm on HPC2N dataset. Thus, the suggested HGHHM algorithm's simulated makespan and resource utilization outperformed previous algorithms, in our experiments, we utilized datasets of varying sizes from 500 to 4000. Conclusions: By using the HGHHM algorithm, this research improves cloud job scheduling efficiency and reliability by improving makespan and resource consumption. These findings confirm hybrid meta-heuristic techniques' efficacy and emphasize the need to balance exploitation and exploration to avoid local optima entrapment. Nevertheless, the study is limited in its scope as it does not take into account other factors such as energy consumption and cost.
Keywords:	Cloud Computing; Henry Gas Solubility Optimization; Harris Hawks Optimization; Task Scheduling.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	GENERATIVE AI: TWO LAYER OPTIMIZATION TECHNIQUE FOR POWER SOURCE RELIABILITY AND VOLTAGE STABILITY
Author:	DR. SANDEEP GUPTA, RAMAKRISHNA KOLIKIPOGU, VEERA SWAMY PITTALA, S.SIVAKUMAR, RAMESH BABU PITTALA, DR. MOHAMMED SALEH AL ANSARI
Abstract:	Wind and solar power are essential in the fight against climate change and for reaching carbon neutrality targets. Due to their inherent unpredictability, renewable energy sources pose a threat to the power system's transient voltage stability, dependability, and flexibility. These consequences might add complexity to power scheme design. This paper introduces a two-tiered optimization approach for control foundations and network design to discourse the effects of renewable vigor on electrical organization preparation, particularly regarding dependability and transient voltage stability. Constructing designs for generators and energy storage units are determined by upper-layer network planning, which assesses the system dependability index. Transient stability demands and constructing and maintenance expenses are addressed by lower-layer challenges. It is suggested to use a two-layer iterative technique using adaptive particle swarm optimization (PSO) for successful nonlinear problem solution. Implementing the suggested approach on an IEEE 33 system of testing demonstrates its practicality. In addition to improving the network's operational efficiency and reliability, the findings show that the suggested optimization method also fixes the issue with regard to program and group planning. Future plans for the operation and planning of the power scheme may be informed by the outcomes.
Keywords:	Climate Change, Renewable Energy, Two-layer Optimization, Voltage Stability, PSO
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	THE IMPACT OF QUANTUM COMPUTING (ALGORITHMS) ON CONTEMPORARY SECURITY SYSTEMS AND FUTURE
Author:	AKKU KUBIGENOVA, ALIMBUBI AKTAYEVA, GALIYA YESMAGAMBETOVA, ALTYNBEK SHARIPBAY, VLADIMIR SUKHOMLIN, NAZERKE AUSSILOVA
Abstract:	In the course of technological evolution, a new scientific point of view arose, reviving interest in the theoretical foundations of quantum mechanics and many new issues combining physics, computer science, and information theory. There is a growing understanding that quantum computation may be a more natural model of computation than the classical model and that fundamental information security issues may be more readily revealed through the concepts of quantum computation. This research aims to contribute to these advancements in these areas: the quantum computer based on the theory of ternary logic is an efficient general-purpose computing device capable of simulating any computing process, increasing the computation time by a polynomial factor only. Quantum computation has the potential to significantly enhance our understanding of security issues, making our work more important than ever. The authors suggest ways to develop optimally using the theoretical principles of the functioning of a quantum computer based on three-digit logic and how to implement structural elements' security using ternary technology.
Keywords:	Three - digit logic, Quantum computing, Quantum theory, Qubits; Cybersecurity.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	A NOVEL ASSOCIATION-BASED ENSEMBLE CLUSTER FOR HEART PREDICTION BASED ON ECG GRAPHS
Author:	G BALA KRISHNA, SIRISHA K L S, V PRADEEP KUMAR, MUMMADI RAMACHANDRA, D SANDHYA RANI, K GNANESHWAR
Abstract:	Traditional image processing techniques usually process data frame-by-frame, which may lead to a loss of temporal information. Temporal aspects are crucial in understanding the ECG waveform's sequential events and dynamic changes. Social Network Analysis (SNA), or graph theory analysis, studies relationships and interactions between individuals or entities within a social network. This analysis can provide valuable insights into various social structures, dynamics, and behaviors. Analyzing electrocardiogram (ECG) data using social graphs is an innovative approach that can provide insights into the cardiac system's dynamics and relationships between different physiological parameters. Although ECG is traditionally used to study the heart's electrical activity, incorporating social graph analysis can offer new perspectives on the interactions and dependencies between various ECG components and their influence on overall cardiac health. Using clustering techniques in social graph analysis can offer several advantages, as it helps identify patterns, groups, and structures within the network. Social networks often consist of diverse data types, such as textual content, user attributes, and interaction patterns. Ensemble clustering can effectively integrate information from different data sources and algorithms, enabling a more comprehensive analysis. The proposed methodology integrates DBSCAN with OPTICS and produces a consensus matrix with the help of an association matrix. DBSCAN and OPTICS are density-based clustering algorithms, but they have different characteristics. DBSCAN is more efficient and straightforward to implement. At the same time, OPTICS provides a hierarchical clustering structure and is more suitable for handling varying-density clusters. Integrating these two algorithms can enhance the overall clustering performance, especially when dealing with complex and varied density datasets
Keywords:	Density-based algorithms, Social Analysis, Ensemble Clustering, Association Matrix, Community Detection
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	A MODEL-DRIVEN APPROACH TO TRANSFORM UML MODELS INTO MONGODB SCHEMAS USING QVTO: FROM PIM TO PSM
Author:	HAMZA NATEK, AZIZ SRAI, ABDELMOUNAIM BADAOUI, FATIMA GUEROAUTE
Abstract:	Translating UML models into efficient NoSQL databases is a complex task within the domain of software engineering. This study addresses the problem of transforming UML models into MongoDB collections using the Model-Driven Architecture (MDA) approach. The research method involves defining metamodels for UML and MongoDB, followed by the development of QVTo transformation scripts to map UML class diagrams to MongoDB document structures. The transformation process was tested on a sample UML model, ensuring the correctness of the generated MongoDB schema. The findings demonstrate that the QVTo transformation scripts can accurately convert UML models into MongoDB collections, preserving the integrity and semantics of the original UML diagrams. The conclusions highlight the effectiveness of the MDA approach in bridging the gap between UML-based designs and NoSQL database implementations, providing a robust solution for data management in complex systems.
Keywords:	Model-Driven Architecture (MDA), UML to NoSQL Transformation, QVTo Transformation Scripts, MongoDB Schema Generation, Automated Data Modeling
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	A NOVEL ADAPTIVE NETWORK INTRUSION PREVENTION SYSTEM FOR INTERNET OF THINGS
Author:	PARTHIBAN ARAVAMUDHAN, DR.T. KANIMOZHI
Abstract:	Data traffic has significantly increased as a result of recent advancements in computer networks and other related services. At the same time, the negative effects of cyber-attacks have become far more prevalent. Variety of network threats which are increasing posing more difficulties and challenges than ever before. To overcome this issue, anomaly prevention and signature-based prevention are two key effective approaches for preventing the cyber attacks. Even though these two key approaches have a potential for success, but still it is like a “Cloud on the Horizon”. Anomaly-based prevention has the potential for high false positives, while Signature-based prevention is more vulnerable to zero-day attacks. To address this concern, this paper explores the importance in developing an Intrusion Prevention System (IPS) model using Deep Learning (DL) algorithm. To evaluate the performance, NIDS dataset V.10 2017 is utilized. The Z-score normalization technique is used to clean the dataset, in which the scaling approach is applied to remove outlier’s. Adaptive Principal Component Analysis (A-PCA) and Linear Discriminant Analysis (LDA) are deployed for dimensionality reduction and class separation optimization, respectively, resulting in a standardized dataset that accelerates prevention efforts by reducing processing latency and enhancing filtering effectiveness. The Whale Optimization Algorithm (WOA) is implemented to find the best optimal values using the “Rule of Convergence” method. The converged data are identified and trained for feature extraction to build an efficient optimized Convolutional Neural Network (CNN) model. “Re-Routing” concept is introduced, which identifies the threshold values during the hyper-parameter tuning. The primary cause for implementing this idea is to reduce the training cost, time consumption and also to enhance the prevention mechanism. The proposed NIPS model is tested and compared with other existing ML and DL methods. The simulation results affirm that the Optimized Rerouted Convolutional Neural Network (OR-CNN) model outperforms the other avant techniques in terms of time consumption, performance and accuracy.
Keywords:	Network Intrusion Prevention System (NIPS), Convolutional Neural Network (CNN), A-PCA, LDA,WOA, Re-Routing.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	A HYBRID ENSEMBLE APPROACH COMBINING BAGGING AND SVM FOR ALZHEIMER'S DISEASE STAGE CLASSIFICATION
Author:	S.CHITHRA, DR.R.VIJAYABHANU
Abstract:	Neurodegenerative illnesses, such as Alzheimer's disease, cause brain cell damage, resulting in structural loss and neuron death, with Alzheimer's being a common form of irreversible dementia in its advanced stages. Researchers are looking into biomarkers, neuroimaging, and machine learning to improve early diagnosis and care of Alzheimer's patients. Effective treatment of Alzheimer's disease (AD) depends on a precise medical diagnosis, and typical protocols involve constructing a single classifier by extracting features from longitudinal MRI data. When tested on the ADNI dataset for older persons, the ensemble bagging SVM model performs better than other approaches, demonstrating greater performance in important evaluation measures like accuracy, sensitivity, precision and recall.
Keywords:	Alzheimer’s disease, SVM, Ensemble, Bagging, MRI.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text

Title:	SKIN CANCER DIAGNOSIS SYSTEM ON OBJECT DETECTION USING VARIOUS CNN YOLOV5 IN ANDROID MOBILE
Author:	EVI FATIMATUR RUSYDIYAH, DIAN CANDRA RINI NOVITASARI, FARIS MUSHLIHUL AMIN, ANISA NUR AZIZAH, HUDA FEBRIANTO NURROHMAN, DINA ZATUSIVA HAQ, ELEN RISWANA SAFILA PUTRI
Abstract:	Skin cancer is a type of cancer that is dangerous and causes death. In Indonesia, skin cancer is one of the most common health problems, with the number of sufferers ranging from 5.9% to 7.8% per year, making it one of the most common cancers besides cervical cancer and breast cancer. This research aims to create a skin cancer diagnosis system based on object detection using deep learning artificial intelligence based on Android mobile. The method used in this research is CNN YOLOv5, which focuses on real-time object detection. In this study, manual annotation was carried out according to 8 classes of skin lesion objects, of which 3 were skin cancers. Several experiments were carried out to obtain the best skin cancer detection system. In the trial, YOLOv5m had a small average evaluation value compared to YOLOv5l and YOLOv5x. However, the computing system performance on YOLOv5m is much lighter and faster than on YOLOv5l and YOLOv5x. The YOLOv5m model produces sensitivity values >80% in almost all classification classes. Therefore, YOLOv5m is the optimal model for an accurate and real-time skin lesion detection system, especially for skin cancer. The implementation on Android phones aims to make it easier for users to detect early and participate in controlling the growth of skin lesions.
Keywords:	Skin Cancer, Diagnosis, Object Detection, YOLOv5, Android Mobile.
Source:	Journal of Theoretical and Applied Information Technology 15^th August 2024 -- Vol. 102. No. 15-- 2024
Full Text