		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
July 2025 | Vol. 103 No.13

Title:	OPTIMIZING EMERGENCY RESPONSES VIA DISASTER LSTM ALGORITHM FOR SOCIAL MEDIA ANALYSIS
Author:	S. BABY SUDHA, Dr. S. DHANALAKSHMI
Abstract:	The increasing frequency of natural disasters demands efficient and accurate classification systems for real-time information dissemination. This study develops and evaluates classification systems for tweets related to six disaster types—hurricane, forest fire, tornado, drought, flood, and fire—aiming to enhance disaster response capabilities through improved classification accuracy and sentiment analysis. The study uses a dataset of labeled tweets, incorporating both disaster type and sentiment labels (positive, negative, neutral) to assess public reactions. By integrating sentiment analysis into the classification process, emergency responders can identify disaster events and gauge the emotional tone of the population in real-time, providing crucial insights for better decision-making. Two models were implemented: a K-Nearest Neighbors (KNN) algorithm and a Disaster Long Short-Term Memory (DLSTM) network. The KNN algorithm, used as a baseline, showed adequate performance but struggled with the complexities of sequential text data and sentiment analysis. In contrast, the Disaster LSTM model, which utilized both disaster classification and sentiment analysis, significantly outperformed KNN. It achieved an overall accuracy of 84.0%, with 87.2% for hurricanes and 81.9% for droughts, while also demonstrating strong performance in sentiment classification. The novelty of this study lies in its dual-focus model, which simultaneously classifies disaster types and associated sentiments using a fine-tuned LSTM network. This integrated approach offers a more comprehensive understanding of real-time public discourse during disasters—unlike prior studies that addressed these aspects in isolation. These improvements enable faster disaster identification, more efficient resource allocation, and a better understanding of public emotions during crises. This research contributes to the fields of natural language processing and disaster management by constructing a robust framework for leveraging social media data for real-time situational awareness and emergency response.
Keywords:	Disaster Response, Disaster Long Short Term Memory, Tweet Classification, Real-Time Information, Natural Language Processing, Deep Learning, Social Media Data.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	A NOVEL METHOD TO ESTIMATE HIDDEN NEURONS IN ENSEMBLE FLOOD FORECASTING MODEL
Author:	NAZLI MOHD KHAIRUDIN , NORWATI MUSTAPHA2, TEH NORANIS MOHD ARIS, MASLINA ZOLKEPLI
Abstract:	Machine learning model such as neural networks have been widely adopted to provide flood forecast. Self-adaptability in neural networks enable them to learn pattern by their own and adjusted the connection between the neurons, but still producing error. In many neural network adaptations for flood forecasting, the number of hidden neurons is normally randomly selected which can caused overfitting problem in the network. In this study, a novel method to estimates the hidden neuron is proposed to overcome this problem. This method integrates the evaluation of various convergence theorem criteria with grid search to estimates the hidden neuron. By having this integration, optimal number of fix hidden neurons can be determined. This method is used in the ensemble model that based on neural networks to forecast the water level based on rainfall data. Based on the performance measurement using Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Nash Sutcliffe Efficiency (NSE), it is found that the integration of convergence theorem and grid search can be used to fix the number of hidden neuron and has reduce the error that led to overfitting in the forecast.
Keywords:	Neural network, Hidden neuron estimation, Flood forecasting, Grid search, Convergence Theorem
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	ENHANCED ENERGY EFFICIENCY AND POWER ANALYSIS IN SOFTWARE-DEFINED NETWORKS THROUGH MALICIOUS SWITCH DETECTION
Author:	Dr. THANGARAJ E , Dr. MOHAMED MALICK , Dr. AROCKIA JAYADHAS S, Dr.BARKATHULLA , Dr. HARIHARASUDHAN S, R MATHU SUDHANAN
Abstract:	Software Defined Networking (SDN) has progressed by a network model to assist with different requirement of a real time traffic flow in a huge scale switch. The basic function of SDN plays a vital role in partition of data plane from the controller plane with accepting the needed changes as per the necessities. SDN provides the broad programmability of switches and its architecture concepts for the services and applications. In our proposed system, deep learning atmosphere is generated using TensorFlow and Keras atmosphere. It involves in Kaggle dataset for the trusted and the malicious flow detection. Through analysis part, power analysis is added in the controller built. SDN overall performance depends on energy consumption, load balancing and traffic management. In this proposed approach energy efficiency and badge power of the Software Defined Network is presented. We focused on the energy efficiency of the network by improving the already proposed route selection algorithm. We proposed a method for the analysis to determine the power management functionalities in various domains. The stimulation results show the power analysis feature classification helps to identify the trusted flow of switches. We propose an empirical algorithm and mathematical model of energy consummation determines controller plane to maximize the network performance and scalability. The overall performance study determines that the proposed system of drop rate, constant throughput, load balancing and energy efficiency of switches in SDN is better than the already existing algorithms. The comparative study determines the proposed algorithm is approximately saves 30% of energy.
Keywords:	Software Defined Networking, Kaggle, Convolutional Neural Networks, Keras, Power Analysis, load balancing, energy consumption, Tenser-flow.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	ENHANCING BRAIN TUMOR CLASSIFICATION THROUGH ADVANCED IMAGE PROCESSING, HYBRID FEATURE EXTRACTION, MRMR FEATURE SELECTION, AND MULTI-CLASS CLASSIFIERS
Author:	GOURI SANKAR NAYAK, DR. PRADEEP KUMAR MALLICK, DR. BHUSHAN MARUTIRAO NANCHE, DR. NEELMADHAB PRADHI, DR. DILLIP RANJAN NAYAK
Abstract:	The purpose of this research is to create a holistic methodology for detecting and classifying brain tumors using up to date image processing, feature extraction and machine learning methods. To improve the diagnostic accuracy, this research evaluated the model performance of various classifiers as well as expanding feature representation to select the best available model for Brain tumor classification. Techniques: Such process includes developing techniques of image enhancement using TV-L1 norm and MRI segmentation algorithms etc. Preparation: Using sophisticated MRI segmentation techniques, the areas of tumors were precisely identified and improved image quality by implementing the TV-L1 standard for efficient image augmentation. For robust feature extraction and feature richness, Multi Scale Local Binary Pattern (MSLBP) for texture analysis was integrated with the multi-dimensional feature representation feature through the use of Quaternion Wavelet Transform (QWT). The Histogram of Oriented Gradients (HOG) is leveraged to quickly capture edge and shape data. Selecting Features Effectively: the dimensionality is reduced and hence increased classification effectiveness by selecting and prioritizing the most useful features via the Minimum Redundancy Maximum Relevance (MRMR) technique. Different Classification Techniques: Random Forest has great accuracy and is robust, so a number of classifiers were constructed. When it comes to modeling complex data distributions, Support Vector Machines (SVMs) prove effective. The K-Nearest Neighbors (KNN) is utilized for local patterns because it is user friendly and good. Results: The dataset from China's Nanfang Hospital and General Hospital was used to evaluate the techniques and is available on Kaggle. With low error rate (0.0147), high sensitivity (98.53%), and specificity (99.51%), Random Forest (RF) had the highest accuracy of 98.53%. In the end, Naive Bayes (NB) performed the worst at 96.67% while SVM and KNN produced slightly lower accuracies at 97.22% and 97.50% respectively. More sophisticated processing, feature extraction, and selection procedures used as part of the suite of preprocessing operations contributed significantly to the increased classification accuracy. RF revealed great promise for clinical use in identifying brain tumors, achieving high sensitivity and specificity while decreasing false positives to an accuracy rate of 98.53%. Even though RF consistently outperformed the others like SVM and KNN on all measures, they did very well on their own as well.
Keywords:	Histogram of Oriented Gradients, K-Nearest Neighbors, Multi-Scale Local Binary Pattern, Naïve Bayes Classifier, Quaternion Wavelet Transform, Random Forest Classifier, Support Vector Machine, TV-L1 norm
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	A FUTURE TREND ON 5G NETWORK SUB-SLICING TECHNIQUES FOR MACHINE LEARNING ALGORITHMS
Author:	A. KARTHIKA, J. YAMUNA BEE, S. ARUN KUMAR, Dr. A. ANNA LAKSHMI, Dr. G. UMA MAHESWARI, Dr. NAGARAJAN GURUSAMY, N. RAGAVENDRAN, Dr. M. VARGHEESE
Abstract:	One of the primary objectives of 5G networks is to meet the need for vertical services. Artificial reality/virtual reality, electronic health records, live video streaming, robotics, driverless vehicles, and many other applications might benefit from 5G networks. Optimal 5G Network Sub-Slicing Automation (ONSSA) is a novel machine learning framework introduced in this paper for autonomous and dynamic 5G network slice processing. After analyzing historical data and the present state of the network, the framework uses the LazyPredict module to automatically choose the most effective unsupervised learning algorithms. For systems that run concurrently at numerous infrastructure layer levels, network slicing poses very difficult and critical security challenges. Some security challenges that need attention are as follows: It is possible to implement security-based interslice, interslice, and multidomain. The data is vulnerable to hacking in situations when a conventional network based on machine learning is used. Accordingly, Data Communication Network as a Service (DCNaaS), is a whole new service model for Mobile Network Organizations (MNOs). To evaluate the framework, we used Python for dynamic testing and Anaconda Spyder for machine learning implementation. To address the slice allocation problem, this research makes use of datasets and machine learning techniques. We have evaluated the slice allocation approach using many ML models. We demonstrated in the simulation that the proposed method correctly allocates the best slice for service.
Keywords:	5G Networks, Optimal 5G Network Sub-Slicing Automation, Lazypredict, Mobile Network Organizations, Data Communication Network As a Service, Machine Learning
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	EARLY PREDICTION OF HEART DISEASE USING DEEP BELIEF-ASSISTED NEURAL PREDICTOR (DBANP) MODEL
Author:	M.RANJANI, DR.P.R.TAMILSELVI
Abstract:	Heart Disease is considered to be the one of the major diseas.Nowadays it makes the loss of numerous lives in our country. For this we need to predict the diseases. The Deep Belief-Assisted Neural Predictor (DBANP) model is proposed in this study to increase the accuracy of heart disease prediction. The model overcomes the drawbacks of previous models like Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM), which frequently suffer from vanishing gradients and lengthy training times, by combining an ANN for initial feature extraction with a DBN for capturing intricate hierarchical dependencies. Models were developed and evaluated using the Cleveland Heart Disease Dataset and the Cardiovascular Disease Dataset. To guarantee high-quality input data, extensive pre-processing was carried out, including managing missing values and feature selection. Using important measures including Accuracy, Precision, Recall, Specificity, F1 Score, AUC-ROC, and AUC-PR, the DBANP model was contrasted with RNN and LSTM models. According to experimental data, the suggested model performs noticeably better than current models, providing increased predictive power and resilience. By offering a hybrid deep learning system for early disease identification that is both scalable and flexible, this study advances predictive healthcare.
Keywords:	Heart disease prediction, DBANP model, artificial neural network, deep belief network, Cleveland Heart Disease Dataset, Cardiovascular Disease Dataset, deep learning, early diagnosis, predictive healthcare.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	POSTQUANTUM MERKLE SIGNATURE BASED ON MODIFIED LAMPORT ALGORITHM
Author:	LARISA CHERCKESOVA, ELENA REVYAKINA, NIKITA LYASHENKO
Abstract:	With the advancement of quantum computing, conventional digital signature algorithms such as RSA and El-Gamal are increasingly vulnerable to quantum attacks. This presents a significant challenge for ensuring the security and integrity of electronic signatures. As a result, there is a need to develop post-quantum signature algorithms that can withstand attacks from quantum computers while maintaining computational efficiency. This study proposes a novel modification of the Merkle post-quantum signature scheme, integrating an optimized version of Lamport’s one-time signature algorithm. The core contribution of this work is the design and implementation of a new algorithm that significantly reduces signature verification time without compromising cryptographic strength. A software implementation of the modified scheme was developed using Python and the PyQt 6 library, allowing for practical testing and analysis. The performance of the modified algorithm was compared against the standard Lamport algorithm using execution time measurements and statistical analysis. The experimental results demonstrate that the modified algorithm significantly improves signature verification speed, reducing the time required by up to 44.81% compared to the standard algorithm. The study presents a more efficient post-quantum Merkle signature scheme with a modified Lamport algorithm that enhances signature verification speed while maintaining strong cryptographic security. The results suggest that the proposed scheme is particularly well-suited for environments where fast authentication of multiple digital signatures is required. Experimental results confirm the advantage of the proposed approach, offering a more efficient solution for secure, high-speed digital authentication in post-quantum environments.
Keywords:	Post-Quantum Algorithm, Digital Signature, Merkle Signature Scheme, Lamport Signature.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	DRAGONFLY-INSPIRED OLSR PROTOCOL FOR INTERFERENCE-AWARE SEAMLESS CONNECTIVITY IN FANET
Author:	GUNAVATHY H , Dr. RAVINDRANATH P.V
Abstract:	Flying Ad Hoc Networks (FANETs) are vital for various applications, relying on unmanned aerial vehicles (UAVs) for communication in dynamic and challenging environments. Seamless connectivity within FANETs is crucial for uninterrupted data exchange and mission success. However, interference from terrain obstacles, weather conditions, and other UAVs poses significant challenges to routing efficiency. The proposed “Dragonfly-Inspired OLSR Protocol (DO-OLSR)” introduces a novel approach by integrating dragonfly-inspired optimization techniques into the OLSR protocol. This integration minimizes control overhead and route discovery latency, optimizing network performance. The protocol incorporates interference-aware mechanisms that dynamically adapt routing decisions based on real-time environmental conditions, mitigating interference effects. Through simulation-based evaluations, the protocol demonstrates improved network performance, reduced packet loss, and enhanced throughput compared to traditional routing protocols. By dynamically adapting to real-time environmental conditions, the DO-OLSR maintains seamless connectivity while mitigating interference effects, showcasing its potential to enhance overall network reliability and performance in FANETs.
Keywords:	FANET, UAV, Seamless Connectivity, Interference Mitigation, Dragonfly Optimization, OLSR Protocol
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	ENHANCE FAKE REVIEW DETECTION: A HYBRID APPROACH OF IMPLICIT ABSA AND IMBALANCED DATASET HANDLING
Author:	LEENA ARDINI ABDUL RAHIM, KHYRINA AIRIN FARIZA ABU SAMAH, NOOR HASIMAH IBRAHIM TEO, ANIS AMILAH SHARI
Abstract:	Online shoppings convenience has increased reliance on reviews, but fake reviews undermine trust in e-commerce. Many detection models analyze full review text, while overlooking subtle cues like lack of specificity, repetitive wording, and exaggerated sentiments expressed implicitly. Additionally, where genuine reviews far outnumber fake ones, dataset imbalance leads to biased machine learning models with reduced reliability. To address these challenges, this study introduces a hybrid approach integrating Implicit Aspect-Based Sentiment Analysis (ABSA) using Bidirectional Encoder Representations from Transformers (BERT) for implicit aspect extraction and Sentiment Analysis with the Synthetic Minority Over-sampling Technique (SMOTE) for handling imbalanced data. Rule-based indicators identify fake reviews, while a Support Vector Machine (SVM) with k-fold cross-validation evaluates performance. The dataset, sourced from Kaggle’s Amazon Reviews, contains 2,852 reviews across four product categories: foods, home care, personal care, and refreshments. Before SMOTE, the average k-fold recall was 78%. After SMOTE, it rose to 95%, enhancing the approach’s ability to detect most fake reviews despite some false positives. The final result of the constructed hybrid approach achieved 96% accuracy, 60% precision, 100% recall, and a 75% F1 score. We evaluate the performance against two comparative feature approaches, which are (i) an SVM baseline and (ii) a BERT + Rule-based + SVM without SMOTE. High recall ensures effective fake review detection, though lower precision results in some false positives. It concludes that this study enhances trust in online reviews and supports informed purchasing decisions. Future research should expand labelled datasets and explore alternative techniques like Edited Nearest Neighbors to refine the precision-recall trade-off.
Keywords:	Fake Review Detection, Sentiment Analysis, Aspect-Based Sentiment Analysis, Implicit ABSA, Imbalanced Dataset Handling, BERT, SMOTE
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	AI-POWERED SPECULAR IMAGE ANALYSIS FOR CORNEAL ENDOTHELIUM DYSTROPHY PROFILING
Author:	KAMIREDDY VIJAY CHANDRA, E.K.MOUNIKA, SWATHI SAMBANGI, D.MANJU , POORNAIAH BILLA, SYEDA SADIA FATIMA, N. SYAMALA
Abstract:	Human cornea consists of ‘5’ layers the thickest layer is stroma layer and densest layer is the endothelium layer, the hexagonal structured cells is occupied in the endothelium layer, The cell gets disturbed when dystrophies like fuch’s dystrophy (FD), advanced fuch’s dystrophy (AFD), posterior polymorphous corneal dystrophy, Irido corneal dystrophy (ICD), mild polymegathism, and corneal guttata (CG) encounters..Total ‘13’ images are acquired from specular microscope of different dystrophies from various patients and processed into the artificial intelligent convolution filter (AICF) algorithm, which extracts mean cell area of endothelium cell, elongation of endothelial cell, Heywood circularity of endothelial, and compactness of endothelial, hexagonality, standard deviation, co-efficient of variation of endothelium layer (layer 5). The mean cell area of images I1 to I13 varies in the range of 79.5 µm2 to 3485 µm2, elongation endothelial cell varies in the range of 3.11 µm2 to 3.98 µm2, compactness factor of endothelial cell ranges 0.62 µm2 to 0.93 µm2 and Heywood circularity factor ranges 0.8 µm2 to 1.98 µm2, coefficient of variation ranges 10 µm2 to 99 µm2 and hexagonality of endothelial cell 46.6 µm2 to 65.2 µm2 are meticulously calculated. These endothelial statistical parameters represent the healthy condition of endothelial layer.
Keywords:	Fuchs Dystrophy, Corneal Guttata, Heywood Circularity Factor, Endothelium Layer, Compactness Factor
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	IMPACT OF MINIMUM SPANNING TREE ALGORITHMS ON EXTRACTIVE ARABIC TEXT SUMMARIZATION APPROACH
Author:	AKRAM EL KHATIB , GAMAL BEHERY , REDA ELBAROUGY
Abstract:	The purpose of this research is to investigate the impact of using Minimum Spanning Tree (MST) algorithms for enhancing the Arabic Text Summarization (ATS) graph-based approach's performance. The previous researches were conducted in an extractive ATS that relied on a graph approach are very limited, and their performance is still low. This low performance is attributed to the characteristics of Arabic language which is morphologically complex, moreover, there is a lack in ATS researches using graph-based technique. The final results of the graph-based technique mainly rely on the weights between sentences as major features which are poorly calculated. To address these limitations, this study applies and evaluates three MST algorithms (Prim’s, Kruskal’s, and Bourka's) within a single-document extractive ATS system. The proposed system converts text into a graph where sentences are nodes and similarity-based weights are used as edges. The MST algorithm is then applied to extract the most representative sentences. To ensure objective comparison, the Essex Arabic Summaries Corpus (EASC) was used as a benchmark dataset. Experimental results show that Kruskal’s MST algorithm achieves the best performance, demonstrating a significant improvement of 15.2% in recall and 14.3% in F-measure over previous single-document extractive ATS methods. This confirms the effectiveness of MST-based graph algorithms in improving Arabic text summarization quality.
Keywords:	Extractive Arabic Text Summarization, Arabic NLP, Graph Model, Minimum Spanning Tree Algorithm.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	REAL-TIME RESPIRATORY SOUND CLASSIFICATION FOR REMOTE DIAGNOSTIC SYSTEMS UTILIZING DEEP LEARNING AND SPECTRUM ANALYSIS
Author:	MADHAVI LATHA PANDALA, B. VISHNU VARSHAN, K. SNEHITH, Y SUMANTH, KANCHARLA PRAVEEN KUMAR, G N SOWJANYA
Abstract:	The detection of abnormal lung sounds is a critical issue that falls under the diagnosis of respiratory conditions and holds promising developments in deep learning to address such issues. This study proposes a new methodology for classifying adventitious RS by using a remote stethoscope vest coat installed with deep CNNs. The process begins with preprocessing the raw audio data into standard waveforms, transforming the latter waveforms to spectrograms for further processing. Another Fourier Transform is carried out on the data to extract its frequency features that aid in improving discriminative patterns identification in lung sounds. Horizontal flipping is among the techniques for augmenting the data to avoid overfitting. Classifiers such as VGG, AlexNet, ResNet, Inception Net, and LeNet were tested for their classification performance in respiratory sound spectrograms. From all the models built with VGG, VGG-B1 proved to have the highest precision, recall, and accuracy values (96%). There are four types of aberrant lung sounds in the dataset: wheeze, rhonchi, stridor, and crackles, which are obtained from R.A.L.E. Lung Sounds and Easy Auscultation. The proposed system, therefore, provides an efficient and robust solution in real-time detection and classification of abnormal lung sounds as a step toward remote monitoring for early diagnosis of respiratory disorders.
Keywords:	Abnormal Lung Sounds, Deep Convolutional Neural Networks (CNNs), Remote Stethoscope Vest Coat, Respiratory Sound Classification, Spectrogram, Fourier Transform, Data Augmentation.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	MULTI-AGENT PATH PLANNING BASED ON IMPROVED ASYNCHRONOUS ACTOR-CRITIC AGENT ALGORITHM
Author:	TEH NORANIS BINTI MOHD ARIS , CHEN NINGNING , NORWATI MUSTAPHA , MASLINA ZOLKEPLI
Abstract:	Existing approaches often rely on centralized or reactive planning methods, which can become inefficient or cause deadlocks in complex environments. There is a clear need for a process that handles these challenges while maintaining real-time performance and robustness. We propose a multi-agent path planning method based on the improved Asynchronous Actor-Critic Agent (A3C) algorithm. First, we enhance the Actor-Critic neural network within the A3C framework by integrating it with the VGG network to develop a fully decentralized strategy. Additionally, we improved the reinforcement learning rewards and penalties, allowing agents to perform real-time reactive path planning with partially visible information, exhibiting implicit coordination. Second, we use the sequence information input characteristics and long-term memory capabilities of Long Short Term Memory (LSTM) neural networks, allowing the neural network to gain "long-term memory" and enhance network learning speed. The LSTM neural network replaces the deep neural networks (DNN) in the original A3C network. We tested this approach in various environments with different sizes, agent quantities, and obstacle densities. Finally, we quantified the results based on average planning length, average planning time, and success rates over 100 tests. The experimental results show that the proposed method significantly improves the success rate and efficiency of multi-agent path planning in noisy and uncertain environments.
Keywords:	Multi-Agent Path planning, A3C, VGGnet, LSTM
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	DESIGN AND EVALUATION OF HYBRID EXPLAINABLE AI INTERFACES FOR VISION RESTORATION IN NEXT-GENERATION RETINAL PROSTHETICS
Author:	SHANMUGA SUNDARI M , VIJAYA CHANDRA JADALA , SIREESHA VIKKURTY
Abstract:	The development of retinal prosthetics has advanced significantly in recent years, yet challenges remain in achieving both high-quality vision restoration and interpretability of prosthetic function. This paper presents a novel framework—Bio-Optical Explainable Interfaces (BOEI)—which integrates biological signal modeling, optical encoding, and explainable artificial intelligence (XAI) to enhance both the efficacy and transparency of retinal prosthetic systems. BOEI employs a hybrid AI approach combining physics-informed neural networks with interpretable deep learning modules to translate visual stimuli into neural signals tailored for the damaged retina. The system models retinal ganglion cell responses while incorporating feedback loops that visualize and explain the decision-making processes of the AI components. Benchmarked against existing retinal interface models, BOEI demonstrates improved reconstruction accuracy (up to 27% over baseline models) and offers interpretable visual heatmaps correlating prosthetic output with retinal anatomy and cognitive perception metrics. The proposed framework represents a critical step toward clinically viable, trustworthy, and adaptive retinal prosthetics that align with both biological plausibility and patient-specific needs.
Keywords:	Retinal Prosthetics, Explainable Artificial Intelligence (XAI), Bio-Optical Signal Modeling, Eye Floaters Impact Analysis, Neural Signal Prediction
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	ADVANCED IMAGE ANALYSIS OF CORNEAL ENDOTHELIAL DYSTROPHIES USING ARTIFICIAL INTELLIGENT CONVOLUTION FILTERS
Author:	KAMIREDDY VIJAY CHANDRA , A.JOY PRANAHITHA , D.MANJU , SWATHI SAMBANGI ,POORNAIAH BILLA , P.SAMPURNA LAKSHMI, N SYAMALA
Abstract:	The human cornea is a complex structure composed of five distinct layers, each playing a crucial role in maintaining vision. Among these layers, the stroma stands out as the thickest, while the endothelium layer exhibits the highest cell density. The endothelium layer is characterized by hexagonally structured cells, which normally maintain the cornea's health and function. The major diseases such as Fuch's dystrophy (FD), advanced Fuch's dystrophy (AFD), posterior polymorphous corneal dystrophy, Irido corneal dystrophy (ICD), mild polymegathism, and corneal guttata (CG) plays crucial role in reduction of endothelium cells in cornea. To know the condition of endothelium cells we conducted a study using a specular microscope, capturing a total of 13 images from different patients, each exhibiting FD, AFD, ICD, CG etc., To analyze these images, we utilize the Artificial Intelligent Convolution Filter (AICF) algorithm, which enabled us to extract key parameters related to the endothelium layer. These parameters are instrumental in assessing the health of the corneal endothelium and understanding how different dystrophies affect cell morphology. The parameters extracted by the AICF algorithm include Mean Cell Area (MCA), This parameter quantifies the average cell size, with values ranging from 79.5 µm² to 3485 µm². Elongation of Endothelial Cells (EEC) measures the shape of endothelial cells and varies from 3.11 µm² to 3.98 µm². Compactness Factor (CF) reflects the closeness of endothelial cells, with values spanning from 0.62 µm² to 0.93 µm². Heywood Circularity Factor (HCF) assesses the roundness of endothelial cells and exhibits a range of 0.8 µm² to 1.98 µm². Coefficient of Variation (CV) provides insights into the variability in cell sizes, with values ranging from 10 µm² to 99 µm². Hexagonality of Endothelial Cells (HEC) indicates the regularity of cell shapes and varies from 46.6 µm² to 65.2 µm².The statistical parameters serve as valuable indicators of the overall health of the endothelial layer and offer significant insights into how different corneal dystrophies impact the morphology of these crucial cells. By studying these parameters, we can better understand the progression and effects of various corneal conditions on the corneal endothelium. The AICF algorithm efficiently extracts the clinical features with diagnostic quality results 1000 milliseconds, which enhances the clinical interpretation time by over 90%, The algorithm offers scalable and reliable solution for automated corneal image analysis.
Keywords:	Fuchs Dystrophy, Advanced Fuch's Dystrophy, Corneal Guttata, Artificial Intelligent Convolution Filter, Elongation Of Endothelial Cells, Compactness Factor.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	REFINEMENT OF A SOFT-CORE PROCESSOR IMPLEMENTATION ON FPGA
Author:	GYOO SOO CHAE
Abstract:	CPUs with soft cores represent a category of microprocessors whose architecture and functionality can be delineated entirely utilizing languages used to describe hardware, such as Verilog or VHDL. These processors offer a high degree of customization tailored to specific applications and are deployable on platforms for reconfigurable hardware, including Field Programmable Gate Arrays (FPGAs). After the design is successfully implemented, the potential exists for developing Application Specific Integrated Circuits (ASICs) for large-scale production. This study entails designing, simulating, and validating an 8-bit processor utilizing VHDL. The envisaged processors are intended for application in control systems characterized by modest to moderate complexity. Moreover, the 8-bit soft processor design lays the groundwork for potential ASIC development. The simulation outcomes, facilitated by a commercial software (ModelSim), are comprehensively illustrated via timing diagrams. Additionally, the practical outcomes of this design's execution on the VIRTEX-5 are meticulously presented through laboratory experimentation. The primary contribution of this study is the development of a compact, application-focused soft-core processor architecture that emphasizes predictable execution, minimal response time, and efficient use of hardware resources—features that are crucial for embedded control system applications. The simulation and practical implementation results offer insights into the performance and feasibility of deploying such processors in real-world scenarios, paving the way for potential ASIC development and broader industrial application.
Keywords:	Soft-core processor, FPGA, VHDL, Verilog, Hardware description language, ASIC, customization, Reconfigurable hardware, Control applications, Simulation software(ModelSim), implementation, Xilinx Spartan-3, Laboratory experimentation
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	A HYBRID LEARNING APPROACH COMBINING GRAPH AND TRANSFORMER MODELS FOR COMMUNICATION EFFICIENT DISTRIBUTED CONTROL AND ESTIMATION IN NETWORKED SYSTEMS
Author:	GUDURU JAHNAVI , S PRIYANKA
Abstract:	The operation of the Networked Control Systems underpins modern-automated industries including cyber-physical systems, large-scale infrastructure and process applications. However, classical methods of Distributed Model Predictive Control and Event-Triggered Distributed Estimation suffers from inefficiencies such as communication overhead, poor state estimation, slow adaptation to system variations, and short predicting horizons. The methods currently in use fail to capture the spatial and temporal dependencies within the networked subsystems, resulting in suboptimal control decisions and excessive computational. To address this, the proposed Integrated Model for DMPC and ETDE includes the following five enhanced methodologies: (1) Graph Neural Network-Based Predictive Control; (2) Attention-driven Event-Triggered Estimation; (3) Transformer-Based Predictive Observer; (4) Meta-Learning-Based Adaptive Control; and (5) Variational Autoencoder-Based Communication-Efficient Control (VAE-CEC). GNN-PC allows for an efficient approach to modeling interdependencies between subsystems, thus strengthening decentralized control decisions. AET-E employs attention mechanisms to focus on updates from relevant subsystems, therefore preventing unnecessary transmissions. TPO utilizes transformers for accurate predictions of long-range states to adopt resilience against data losses. MLAC guarantees the required robustness under non-stationary conditions by promoting quick adaptation to changing environments through meta-learning. VAE-CEC realizes effective communication by compressing high-dimensional state information at a low cost, which does not affect control performance. The integrated model reported communication overhead saving of 50%, better control adaptive capacity by 40%, improvement in accuracy of state prediction by 35%, and control error reduction by 30%. This proves that the proposed work significantly enhances the efficiency and reliability of DMPC and ETDE methods, thereby making real-time distributed control technology more scalable, adaptive, and resource efficient.
Keywords:	Graph Neural Networks, Event-Triggered Estimation, Distributed Model Predictive Control, Transformer-Based Observer, Communication-Efficient Control, Process control.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	SYNERGIZING MACHINE LEARNING AND QUANTUM ANNEALING IN FRAUD PREVENTION SYSTEMS
Author:	S V N SREENIVASU1, RAVI UYYALA, PRANEETH CHERAKU, GARAPATI SATYANARAYANA MURTHY, M. L. M. PRASAD, MANI MOHAN DUPATY, RAMA KRISHNA PALADUGU
Abstract:	With the increasing prevalence of online transactions, the risk of online fraud has become a major concern for individuals, businesses, and financial institutions. Traditional methods of fraud detection often fall short in addressing the dynamic and evolving nature of fraudulent activities. The escalating threat of online fraud necessitates innovative approaches to enhance the efficacy of fraud detection systems. Using a quantum machine learning (QML) strategy that incorporates Support Vector Machine (SVM) supplemented with quantum annealing solvers, this study has developed and implemented a detection framework. Our evaluation of its detection performance was based on a comparison of the QML application's performance with twelve different machine learning algorithms. This research investigates the fusion of classical machine learning algorithms with quantum annealing solvers as a novel strategy for fortifying online fraud detection. With traditional methods struggling to keep pace with the dynamic nature of fraudulent activities, this paper explores the potential synergy between machine learning and quantum computing to address the evolving challenges in online transactions. Our study aims to demonstrate the feasibility and effectiveness of integrating these technologies, leveraging quantum annealing to optimize the complex decision-making processes inherent in fraud detection. Through an in-depth analysis, we present findings on the performance, speed, and adaptability of the integrated model, showcasing its potential to revolutionize the landscape of online fraud detection and bolster cyber security measures.
Keywords:	Cyber security, Fraud detection, Machine learning, Quantum computing, Support Vector
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	PERFORMANCE CHARACTERIZATION OF CACHE REPLACEMENT STRATEGIES FOR MANAGING SHARED LLC OVER HETEROGENEOUS INTEGRATED CPU-GPU ARCHITECTURE USING MULTI-THREADED WORKLOAD
Author:	PRATAP KESHARI PANDA, BANCHHANIDHI DASH, PRASANT KUMAR PATTNAIK
Abstract:	The CPU is mostly designed for sequential and complex decision-making. it propagates with its instruction processing in a few pipeline stages, including steps for data and instruction fetching from memory. With time, the GPU, with its parallel processing capabilities, came into existence to assist the CPU, mainly for graphics processing. Because of parallel processing and a huge number of thread executions, its data access pattern is quite different from that of CPUs. In its rendering pipelines, the GPU may access varied data streams. The access may vary in semantics in different access patterns. Due to commercial requirements and to minimize data sharing latency between CPU and GPU, integrated processor designs are emerging. these designs accommodate both CPU and GPU in a single chip, sharing common resources. On many occasions, sharing these resources becomes a bottleneck for the whole system, deteriorating overall performance. Last Level Cache sharing among CPU and GPU is a bigger challenge in this design approach. A suitable cache eviction strategy is highly essential to utilize the shared LLC space effectively. Optimized cache with better clock speed can also be considered along with a higher configuration for CPU and GPU to improve the overall performance. Here in our work, we have designed an integrated heterogeneous CPU-GPU model with upgraded configurations to boost performance. Further, we have compared the performance with Alder Lake and Raptor Lake processors, taking them as baselines. We have achieved a speedup improvement of 59.8% and 22.12% compared to Alder Lake and Raptor Lake processors, respectively, taking the geometric mean of eight different sets of workloads. Eight different cache replacement schemes have been configured in MacSim simulator, and the workload from different benchmark suites has been given. Through simulation results, we found an average read miss count improvement of 29.41% and 15.38% over Alder Lake and Raptor Lake, respectively. This may help to IT infrastructure in terms of actuarial science, real time systems etc
Keywords:	Multi-Core, Graphics Processing Unit, Multi-Threading, Benchmark, Cache Replacement Policy, Shared Last Level Cache, Heterogeneous Architecture
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	MALICIOUS DOMAIN DETECTION USING INTEGRATED SUPERVISED AND UNSUPERVISED MACHINE LEARNING APPROACHES
Author:	SRI LAXMI KUNA, PULLURI SRINIVAS RAO, A. LAKSHMANARAO, SOWNDARYA LAHARI CHINTALAPUDI, NALANAGULA HARINI, HARI KRISHNA H
Abstract:	Detecting Domain Generation Algorithms (DGA) is crucial in cybersecurity to identify malicious domain names. While existing studies focus individually on either supervised or unsupervised learning, limited work has explored their integrated use for DGA detection. This paper addresses that gap by combining clustering-derived features with traditional classifiers to enhance detection accuracy. This paper explores an innovative approach for DGA detection utilizing supervised classification and unsupervised clustering techniques. The methodology begins with preprocessing the dataset and extracting relevant features, such as domain names, host information, and subclass labels. Later, feature hashing is utilized for dimensionality reduction, transforming categorical features like domain names, hosts, and subclasses into feature vectors. Advanced clustering methods, including KMeans, Hierarchical Clustering (Agglomerative), and Density-Based Clustering (DBSCAN), are employed to uncover underlying patterns in the data. These techniques aid in identifying distinct groups or clusters within the dataset, potentially assisting in differentiating DGA from legitimate domain names. Later, cluster labels were added as features for final dataset. Subsequently, multiple ML classifiers, including Random Forest, Decision Tree, KNN, SVM, and Logistic Regression, are trained to classify domain names as DGA or non-DGA based on the extracted features. Rigorous experimentation and evaluation assess the performance of each classifier in terms of accuracy and other relevant metrics. This hybrid approach contributes new knowledge on how feature enrichment through clustering can improve model generalization in real-world cyber threat scenarios. The results offer insights into the effectiveness of the proposed methodologies for DGA detection.
Keywords:	DGA Detection, Supervised learning, Unsupervised Learning, Machine Learning, Random Forest.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	PREDICTION THE DEPTH OF COLOR, OF CATIONISED COTTON FABRIC DYED WITH REACTIVE DYES USING FUZZY LOGIC
Author:	M. ELKHAOUDII, M. EL BAKKALI, R. MESSNAOUI, A. SOULHI≠4, O. CHERKAOUI
Abstract:	Cotton fabrics are commonly dyed using reactive dyes. These dyes typically require substantial amounts of salt or sulphates, which serve to reduce the surface tension between the anionic dyes and the anionic cellulose fiber. This interaction significantly enhances dye bath exhaustion and facilitates the diffusion 0f the dye onto the cotton fibers. However, reactive dyeing of cotton has a considerable environmental impact due to the large volumes of dye effluent containing unfixed dyes and salts. As a result, reactive dyeing is considered 0ne of the most environmentally polluting dyeing methods. Among the alternative approaches gaining increasing attention from researchers is the chemical cationization of cotton, which improves dye affinity without the need for salt. In this study, cotton was cationized using the commercial product CRW, supplied by Impocolor, to achieve an adequate dye uptake for uniform reactive dyeing on the modified cotton substrate. T0 effectively monitor the dyeing process of cationized cotton with the reactive dye Triactive Red S3B, a Datacolor SF 450 spectrocolorimeter was employed to measure the K/S values of various samples. The effects of temperature, dye concentration, sodium carbonate concentration, and processing time on the dyeing performance of cationized cotton were evaluated.
Keywords:	Cationization, Exhaustion, Reactive dye, Cotton, Fuzzy logic, Color strength
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	SECURING CYBER-PHYSICAL SYSTEMS: FLAMINGO SEARCH ALGORITHM OPTIMIZED DEEP LEARNING FOR THREAT DETECTION
Author:	K. SRI VIJAYA, S. SUDESHNA, NARESH KUMAR BHAGAVATHAM, PARASA KONDALA RAO, N. SRIJA, PRAVEENA MANDAPATI, RAMESH ELURI
Abstract:	Threat detection in Cyber-Physical Systems (CPS) is essential to safeguarding the reliability and security of these integrated systems, which interface digital components with the physical world. CPS platforms, common in healthcare, industrial automation, smart cities, and transportation, face vulnerability to various cyber threats. Effective threat detection in CPS involves identifying and mitigating cybersecurity risks, which can otherwise disrupt physical operations, compromise data integrity, and jeopardize safety. Machine Learning (ML) and Deep Learning (DL) techniques are increasingly leveraged for detecting anomalies by modeling the CPS’s normal behaviour and recognizing deviations. This study presents an Automated Threat Detection using the Flamingo Search Algorithm with Optimal Deep Learning (ATD-FSAODL) in CPS environments. Initially, the ATD-FSAODL technique applies Flamingo Search Algorithm (FSA)-based feature subset selection to identify optimal feature sets. The ATD-FSAODL approach utilizes a modified Elman Spike Neural Network (MESNN) for threat recognition and classification, with the Slime Mold Algorithm (SMA) optimizing the MESNN parameters to enhance detection accuracy. Simulation experiments on benchmark databases demonstrate the effectiveness of the ATD-FSAODL technique, achieving a maximum accuracy of 99.58%, precision of 99.58%, recall of 99.58%, F-score of 99.58%, and MCC of 99.16%.
Keywords:	Cyber-physical system, Threat analysis, Industry 4.0, Deep learning, Feature selection
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	INTELLIGENT STEGANOGRAPHY: CNN-BASED DATA HIDING IN EDGE REGIONS AND NON-OVERLAPPING BLOCKS
Author:	ANUSHA REDDY NARA, SAYED SALMA SULTHANA, DASARI ANUSHA, V. CHANDRA KUMAR, KARI VENKATA SUMANTH, PINJARI MASOOM BASHA, NARENDRA BABU PAMULA
Abstract:	Deep learning methods added to steganography have greatly simplified the hiding of data in digital media without anyone noticing. This article discusses a clever steganography technique hiding information in edge areas using Convolutional Neural Networks (CNNs), hence covering picture blocks not overlapping. Carefully selected edge sections feature several varied textures, which helps to conceal injected data and reduce visual distortion. By means of local pixel variation, the proposed CNN-based approach automatically locates the optimal embedding sites. This guarantees great payload capacity without compromising undetectability. Unlike other techniques that depend on manual feature engineering, ours employs deep learning to modify the embedding strength depending on the image material, hence enhancing safety and resilience. Blocks that don't overlap save even more computing power. They let processing occur in real time and lower mistakes that can lead steganography to be found, thus We evaluated the proposed approach on several standard datasets and discovered that, in terms of embedding capacity, visual quality (as assessed by PSNR and SSIM), and resistance to statistical attacks, it performs better than conventional LSB and DCT-based approaches. The technology also performs effectively with various kinds of photos, which implies it can be utilized for digital watermarking and secure interaction.By means of a secure, flexible, high-capacity approach to conceal data, this work bridges the gap between machine learning and steganography. Current CNN-based steganography methods often suffer from low payload capacity, especially when data hiding is limited to edge regions that constitute a very small percentage of the image. Moreover, overlapping block methods compromise robustness and invisibility as they do not adapt to local image complexity and incur redundancy and computational overhead. Furthermore, many existing methods are vulnerable to modern steganalysis tools and are not transferable to different types of images due to limited use of deep features and static embedding strategies. Researchers in the future could investigate how antagonistic training could reduce system susceptibility to sophisticated steganography systems.
Keywords:	Steganography, Convolutional Neural Networks (CNN), Edge-Based Embedding, Non-Overlapping Blocks, Deep Learning, Data Hiding, Image Security, Steganalysis Resistance
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	VGG-DAGSVM FOR MULTI-CLASS APPLE LEAF DISEASE DETECTION: A TRANSFER LEARNING APPROACH
Author:	JOICE SHAKILA.A, Dr.SARAVANAN.S
Abstract:	Apple cultivation is essential in world agriculture because of its high nutritional benefits and economic importance. However, increasing pressures from foliar diseases such as apple scab, cedar apple rust, and black rot seriously threaten yield and quality within UK apple orchards. One strategy is to capitalize on advanced technologies such as computer-aided diagnosis systems for improved detection and classification of diseases. To address this challenge, this paper proposes a generic and automatic solution to recognizing and classifying diseases of apple fruit leaves leveraging the VGG-DAGSVM architecture. The workflow begins with bilateral filtering to mitigate image noise without losing core edge information, and Contrast Limited Adaptive Histogram Equalization (CLACHE) for contrast enhancement. SegNet is used to accurately segment diseased regions, and the VGG-19 deep learning model features extraction. Finally, a Directed Acyclic Graph Support Vector Machine (DAGSVM) is used as the classifier for accurate disease classification. Experimental evaluation uses a publicly available dataset comprising 13,124 apple leaf images across four categories. The proposed model achieves a classification accuracy of 96.50%, precision of 96.05%, sensitivity of 95.92%, specificity of 96.43%, and F-score of 96.45%, outperforming several existing models. These results demonstrate the robustness and applicability of the potential of integrating image processing, deep feature extraction, and intelligent classification systems to support early disease detection and promise sustainable apple farming practices.
Keywords:	Apple leaf, Diseases, Deep Learning, Transfer Learning, Support Vector Machine.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	LEARNED DICOM IMAGE COMPRESSION VIA SUPER-RESOLUTION WITH DILATED RESIDUAL BLOCKS AND PIXEL ATTENTION
Author:	OBADA OTHMAN AGHA, YAHIA FAREED, LOUAY CHACHATI
Abstract:	The digital imaging and communications in medicine (DICOM) standard leads global efforts to advance medical imaging. With the growing interest in this field, especially given the rapid developments in telemedicine applications, achieving efficient compression without losing diagnostic accuracy is critical. Therefore, there is an imperative need to employ an advanced hybrid deep-learning model, specifically designed for these medical images, to outperform current methods. The proposed method tests the development of a non-autoregressive model for parallel pixel prediction, avoiding sequential processing. This scheme uses dilated residual blocks (DRBs) to capture long-range dependencies using fewer blocks. It combines depthwise separable convolution (DSC) layers with the pixel attention mechanism to reduce computational complexity while preserving the diagnostic details in the reconstruction task. The method also leverages a discretized mean and scale Gaussian distribution mixture model to achieve efficiency across various DICOM image types. The results demonstrate that the method achieves higher compression ratios, and improves the bpsp metric by 19.26%, 22.89%, and 23.94%, compared to the best competing methods, for each of magnetic resonance imaging (MRI), computed radiography (CR), and computed tomography (CT) images, respectively. Compared to the leading learning-based methods, the system reduces the mean compression time by [16.96 - 19.02] %. At the same time, the high values of PSNR and SSIM metrics demonstrate the ability to ensure high quality. This approach balances compression efficiency, speed, and diagnostic reliability, enhancing DICOM image processing for telemedicine.
Keywords:	DICOM images, Deep learning, Super-Resolution, Images Compression, DSC layers.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	EXPLORING ENHANCEMENT IN SENTIMENT ANALYSIS USING DESCRIPTIVE-SEMANTIC TECHNIQUES: A SYSTEMATIC LITERATURE REVIEW
Author:	NURAIN BATRISYIA BIDIN, KHYRINA AIRIN FARIZA ABU SAMAH, NOR AFIRDAUS BINTI ZAINAL ABIDIN, LALA SEPTEM RIZA
Abstract:	Sentiment analysis (SA) is a critical component of natural language processing (NLP) that enables the extraction of subjective information from large volumes of textual data. However, achieving high accuracy and contextual understanding remains challenging due to the unstructured nature of text and the complexity of linguistic patterns. To address this issue, this study investigates applying descriptive and semantic analysis (DSA) techniques to improve sentiment reliability and contextual accuracy. This study conducts a systematic literature review (SLR) of 28 publications from 2019 to 2024 using the PRISMA methodology, focusing on works indexed in Electrical and Electronics Engineers (IEEE) and Science Direct (SD). The review identifies Exploratory Data Analysis, Term Frequency-Inverse Document Frequency, and Latent Dirichlet Allocation as the most prevalent DSA techniques to enhance SA. Four thematic areas emerged from the analysis: sector, purpose, algorithm, and method employed. The study concludes that while DSA techniques contribute significantly to addressing contextual and semantic limitations in SA, there remains a need for more integrated approaches to handle complex sentiment structures. The findings highlight key challenges, recent advancements, and directions for future research in this evolving field.
Keywords:
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	HYBRID CNN WITH ATTENTION-BASED FEATURE FUSION FOR LUNG DISEASE DETECTION FROM RESPIRATORY SOUNDS USING CLASS-BALANCED OPTIMIZATION
Author:	Dr. P. GNANASUNDARI, Dr. PRADIP RAM SELOKAR, Dr. S. GOPINATH, V.KIRUTHIKA, B.SUGANTHI, EPHIN M
Abstract:	Lung disease detection using respiratory sounds involves analyzing audio recordings of breath sounds to identify abnormalities associated with conditions such as Chronic Obstructive Pulmonary Disease (COPD), pneumonia, and asthma. However, accurate detection remains challenging due to class imbalance in datasets and the difficulty of extracting discriminative features from complex audio signals. This study proposes a novel Hybrid Convolutional Neural Network with Attention-Based Feature Fusion (HCNN-AFF) to enhance detection accuracy. The model integrates spectrogram-based inputs with Mel-Frequency Cepstral Coefficients (MFCCs), effectively capturing both time-frequency and cepstral information for richer feature representation. To address class imbalance, a combination of data augmentation and a class-balanced loss function is employed, enabling improved learning from minority class samples. Furthermore, the model is optimized using the AdamW optimizer enhanced with a LookAhead mechanism (AWLA), promoting better convergence and generalization. Experimental results demonstrate that the proposed HCNN-AFF method outperforms existing approaches in respiratory sound analysis, achieving superior accuracy and robustness in lung disease detection.
Keywords:	Lung Disease, Respiratory Sounds, Machine Learning, Feature Extraction, Mel-Frequency Cepstral Coefficients, Early Diagnosis..
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	UTILIZING ARTIFICIAL INTELLIGENCE PREDICTIVE MAINTENANCE IN LEAN MANUFACTURING TO BOOST INDUSTRIAL SUSTAINABILITY AND ENERGY EFFICIENCY
Author:	WAHYU SARDJONO, DENY ADITYA PRATAMA
Abstract:	The implementation of Artificial Intelligence (AI)-driven Predictive Maintenance (AI-PdM) within Lean Manufacturing has emerged as a transformative strategy to improve energy efficiency and promote industrial sustainability. AI-PdM supports continuous condition monitoring, advanced predictive analytics, and optimized maintenance planning, which help mitigate unexpected downtimes, enhance equipment longevity, and maximize energy utilization. This research adopts a Systematic Literature Review (SLR) approach to evaluate AI-PdM’s impact on manufacturing efficiency and its correlation with Sustainable Development Goals (SDG) 7 (Affordable and Clean Energy) and SDG 9 (Industry, Innovation, and Infrastructure). The study’s findings highlight AI-PdM’s significant contribution to lowering carbon emissions, increasing operational dependability, and fostering sustainable production methods. Nevertheless, obstacles such as substantial initial capital requirements, a shortage of specialized workforce, and difficulties in integrating with existing systems continue to hinder its widespread adoption. Therefore, this paper advocates for the establishment of standardized guidelines and stronger collaboration among academia, industry stakeholders, and policymakers to accelerate AI-PdM adoption in sustainable manufacturing.
Keywords:	Intelligence, Predictive Maintenance, Lean Manufacturing, Energy Efficiency, Industrial Sustainability, Industry 4.0.
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	EXPLORING KEY INFORMATION REQUIREMENT FOR EFFECTIVE MONITORING OF COURSES AND STUDENT PERFORMANCE: A QUALITATIVE THEMATIC ANALYSIS
Author:	MOHD HAFIZAN MUSA, SAZILAH SALAM ,MOHD ADILI NORASIKIN, NORAFFANDY YAHAYA, RUJIANTO EKO SAPUTRO
Abstract:	Universities serve multiple stakeholders, including students, lecturers, parents, alumni, and regulatory bodies. Among these, students stand out as a primary focus, aligning with the university's core mission to educate. Achieving stakeholder satisfaction necessitates enhanced education quality, encompassing teaching methods, program delivery, and the relevance of course offerings. Students' success in gaining knowledge, skills, and competencies is critical to university effectiveness. In today’s dynamic educational landscape, digital technologies offer new avenues to enrich learning, with Learning Analytics (LA) emerging as a transformative field. The growth of big data has increased the volume and complexity of available educational data, requiring LA systems to evolve accordingly. With the universities maintaining more than one e-learning platform, LA presents a new challenge in integrating this heterogeneous e-learning data. To overcome this problem, graph databases and Resource Description Framework (RDF) ontologies have proven valuable for managing and analysing such extensive data. However, to build Web Ontology Language (OWL) frameworks that effectively support LA, researchers must first identify faculty administrators' and lecturers' specific data needs or key information, ensuring only essential information is taken to be integrated in LA. To gather the key information, interview sessions were conducted with six participants from two institutions, including an e-learning coordinator, lecturers, and faculty administrators as part of the case study. The interview result is crucial for identifying the essential information needed in LA pertaining to course performance and student performance.
Keywords:	Learning Analytics, Student Performance, Course Performance, Information Data Retrieval, e-learning, Heterogeneous Data Retrieval
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	HYBRID DEEP LEARNING INTEGRATION ON ADVERSIAL NETWORK FOR AN ACCURATE AGRICULTURAL SEQUENCE DATA PREDICTION
Author:	ARUMAI RUBAN J, SUNDAR SANTHOSHKUMAR , A. SUMATHI, J.JEGATHESH AMALRAJ, R. BHAGAVATHI LAKSHMI
Abstract:	In the Drip Irrigation system, Water scarcity and inefficient irrigation practices are major challenges that are faced in smart agriculture. To overcome these issues lot of Deep Learning (DL) methods are processed for prediction. Though there are lot of inefficiency and inaccurate in evaluation, an effective prediction is required in modern agriculture. To attain higher accuracy and robust prediction, this work presents an advanced sequence of Generative Adversarial Networks (SeqGAN) to generate and predict agricultural data that is specifically used to optimize irrigation practices and manage water resources effectively. The proposed seqGAN architecture consists of a generator and a discriminator. The generator involves a DL method of Long Short-Term Memory (LSTM) networks to create realistic agricultural text sequences by learning from previous data and it also has controlled variability through noise injection. The discriminator includes a Gated Recurrent Unit (GRU) and a CatBoost classifier to differentiate between real and generated sequences. The CatBoost integration enhances the model's ability to handle categorical data efficiently which enhances the accuracy and robustness of sequence classification. This proposed method is particularly beneficial for agricultural datasets augmenting and effective predictive sequences in agricultural fields. This proposed work not only improves data availability but also supports innovative solutions in agricultural research that ultimately contribute to more sustainable and efficient farming tasks than conventional methods.
Keywords:	Agricultural data, SeqGAN, LSTM Generator, GRU Discriminator, CatBoost classifier, Error Rate Analysis
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	EFFICIENT LLM INFERENCE ON MCP SERVERS: A SCALABLE ARCHITECTURE FOR EDGE-CLOUD AI DEPLOYMENT
Author:	SWAPNA DONEPUDI, U POORNA LAKSHMI, NVS PAVAN KUMAR, S LALITHA, RUHISULTHANA SHAIK, DESHINTA ARRORA DEVI
Abstract:	Organizations need to create deployable LLM models which use open frameworks with privacy-protected standards. Cloud-based inference remains popular yet it forces delays along with resource wastage and exposes security threats. The processing limitations of edge computing create an isolated drawback due to the closeness of data sources. The proposed research presents an edge-cloud joined system with MCP servers that delivers efficient LLM inference workload offloading. Research results showed that system performance delays occurred concurrently with resource consumption measurements and throughput capacity assessments and energetic performance metrics measurements. Through the system analysis tool users can evaluate performance and resource utilization and throughput performance simultaneously. Model predictions through this system achieve accurate results while the system performs at industry standards regarding latency and throughput acceleration. Within the proposed framework researchers established edge-Cloud LLM orchestration capabilities to optimize AI systems deployment in current real-world scenarios.
Keywords:	Resource Efficiency, Wireless, Large Language Models, Cloud
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	OPTIMIZED RELIABILITY FRAMEWORK FOR SERIES-PARALLEL SYSTEMS: STRATEGIC REDUNDANCY ALLOCATION USING HAM AND IPM SOLUTIONS
Author:	SRIDHAR AKIRI, RAMADEVI SURAPATI, SRINIVASA RAO VELAMPUDI, ARUN KUMAR SARIPALLI, BHAVANI KAPU & PHANI BUSHAN RAO PEDDI
Abstract:	When it comes to reliability engineering, the main goal is to make sure that systems and their parts always work as they should within a certain time limit and under certain conditions. Within the domain of reliability theory, system robustness is improved by strategically incorporating redundancy, while simultaneously accounting for constraints such as cost, mass, and spatial requirements in series-parallel architectures. This research focuses on examining how these constraints-specifically weight, volume, dimensional, and spatial factors-influence system reliability enhancement. The investigation is centered on spare parts utilized in a representative Automation Forum setup, where factors like cost, weight, and size are vital to sustaining effective functionality. Unlike electronic systems that may not emphasize these constraints, the Automation Forum includes crucial components such as compressors, condensers, and absorption towers, all of which require thorough reliability evaluation. To address this, the Lagrange multiplier method is employed to design and analyze an integrated redundant reliability model configured in a series-parallel structure. This approach yields continuous-valued outputs for essential variables including component count, individual and stage reliabilities, and total system reliability. To derive feasible and practical integer-based solutions, the study further implements heuristic strategies and integer programming. These techniques significantly enhance the precision and relevance of the reliability analysis conducted.
Keywords:	IRR Model, Series-Parallel Configuration, System Reliability, LAM Approach, HAM Approach, IP Approach
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text

Title:	AI AND SCRIPTWRITING: A NEW THREAT?
Author:	LAURENSIUS VICKY CRISTANTO, EKKY IMANJAYA
Abstract:	As Artificial Intelligence (AI) becomes increasingly inevitable in the film industry, a critical controversy arises: Can AI replace human scriptwriters, or does it merely serve as a tool to assist them? This study addresses the debate through an academic investigation of Sunspring (2016), the first AI-generated film, to map the discourse and offer practical solutions. Using a desk review methodology focused on 2021–2024 literature and critical assessments of Sunspring, our findings highlight an urgent need for scriptwriters to master AI as an assistive tool. This approach positions human writers as creative masterminds who harness technological efficiency while avoiding artistic compromises. The paper suggests a teamwork approach, saying that even though using AI is necessary, it should be done carefully to avoid slowing down creativity and to protect the role of human writers. These insights contribute to the AI-creativity debate and provide pragmatic guidance for industry professionals navigating this transformative era. The study concludes that while AI cannot replace human scriptwriters, it can act as a valuable assistant by enhancing ideation; however, it requires human oversight to maintain coherence, emotional depth, and thematic integrity.
Keywords:	Artificial Intelligence, Scriptwriting, Sunspring, Human scriptwriters, Filmmaking
Source:	Journal of Theoretical and Applied Information Technology 15^th July 2025 -- Vol. 103. No. 13-- 2025
Full Text