IJACSA Vol. 17 Issue 10 (2026)

1

Encrypted Malicious Traffic Detection Fusing Cross-Attention with Lightweight Graph Convolution

Author 1: Xiaogang Yuan Author 2: Minghang Ni Author 3: Dezhi An Author 4: Jianxin Wan

With the widespread adoption of encryption technologies, traditional deep packet inspection methods fail to parse encrypted payloads, making it difficult to effectively identify malicious traffic. To address this, this paper proposes a Dual-Level Cross-Attention Lightweight Graph Convolutional Network (DCA-LGCN). This method employs node-level attention to extract intra-flow features and edge-level… Read full abstract & cite →

Encrypted traffic malicious traffic detection graph neural network attention mechanism

Abstract doi.org/10.14569/IJACSA.2026.0171001 PDF

2

A Session-Bound Attendance Protocol with Rotating QR Codes and Uncertainty-Aware Geofencing: Implementation and Controlled Evaluation

Author 1: Smaila Amoanu Author 2: Bright Wiredu Nuakoh Author 3: Stephen Asiedu

A scanned attendance credential does not, by itself, establish that its holder is eligible or physically present. This study specifies and evaluates a session-bound attendance protocol in Arkos AttendPro, a web-based platform for workforce and event attendance. The protocol combines authenticated rotating QR tokens, session and schedule eligibility checks, valid… Read full abstract & cite →

Attendance management rotating QR code geofencing location uncertainty proxy attendance authentication controlled evaluation

Abstract doi.org/10.14569/IJACSA.2026.0171002 PDF

3

Retinex-Based Traffic Sign Recognition for Robust Performance Under Low-Visibility Conditions

Author 1: Jiyoun Jung Author 2: Kyuchang Chang

With the advancement of autonomous vehicles and Advanced Driver Assistance Systems (ADAS), Traffic Sign Recognition (TSR) systems have become increasingly important. However, adverse weather conditions such as fog, rain, and low illumination significantly degrade model performance, leading to missed detections and false positives. To address this challenge, we propose an… Read full abstract & cite →

RetinaNet retinex traffic sign recognition adverse weather conditions german traffic sign benchmark

Abstract doi.org/10.14569/IJACSA.2026.0171003 PDF

4

The Role of ICT-Based Spatial Design Interfaces as Mediators for Promoting Resource Rediscovery Among Local Residents

Author 1: Taiga Haruta

As depopulation progresses on a global scale, regional circulation-type stamp rallies have been implemented in rural areas as a means to foster civic pride and promote the rediscovery of the potential value of their region. However, traditional paper-based methods impose significant burdens on both organizers and users, presenting challenges for… Read full abstract & cite →

Stamp rallies augmented reality location information image recognition AI spatial design interfaces

Abstract doi.org/10.14569/IJACSA.2026.0171004 PDF

5

When AI Agents See Each Other’s Answers, Their Answers Grow More Alike: A Controlled Study of Multi-Agent LLM Revision

Author 1: Hideki Muto Author 2: Tetsuro Ogi Author 3: Takahiro Yakoh

AI systems can offer several answers when a question has no single correct solution. For people seeking alternatives, those answers need to differ. Multi-agent systems often let agents read one another’s answers during revision. Can agents keep their answers different when they read one another’s work? Here we show that… Read full abstract & cite →

Large language models multi-agent systems interagent communication response diversity response convergence

Abstract doi.org/10.14569/IJACSA.2026.0171005 PDF

6

Artificial Intelligence-Based Ensemble Approach for Medical Decision Support Systems on Lung and Colon Cancer Diagnosis Using Histopathological Images

Author 1: Samah Alzanin Author 2: Wafi Bedewi

Cancer accounts for a vast death rate owing to its aggressiveness, massive latent metastasis, and heterogeneity (which causes resistance against chemotherapy). Lung and colon cancers (LCC) are one of the most dominant varieties of cancer all over the world, which occurs in both females and males. Prompt and precise analysis… Read full abstract & cite →

Lung and colon cancer artificial intelligence histopathological images feature extraction Siamese neural network

Abstract doi.org/10.14569/IJACSA.2026.0171006 PDF

7

A Lightweight Forest Fire Detection Algorithm Based on an Improved YOLOv11

Author 1: Jingzhou Tao Author 2: Zijing Liu Author 3: Ruoxi Lyu Author 4: Jie Li Author 5: Xu Cao

This study presents IWL-YOLOv11 (Improved Wildfire Lightweight YOLOv11), a detector that balances forest fire detection accuracy and parameter count. The model introduces three components. A Dual-Branch Gated Depthwise Convolution Block (DGCB) modifies the C3K2 backbone through depthwise separable convolutions and element-wise interaction between two feature branches. A Hybrid Convolution-Attention Network… Read full abstract & cite →

Object detection YOLOv11 dual-branch gated depthwise separable convolution hybrid convolution-attention network forest fire detection

Abstract doi.org/10.14569/IJACSA.2026.0171007 PDF

8

AI and Generative AI for Recruitment and Talent Management: A PRISMA-Based Systematic Literature Review

Author 1: Abdelkrim SAOUABE Author 2: Doha BANOUI Author 3: Khalid TIZRA Author 4: Merouane MAZAR Author 5: Imad MOURTAJI

The rapid emergence of Artificial Intelligence (AI) and Generative Artificial Intelligence (GenAI) is reshaping Human Resource Management (HRM), transforming traditional recruitment and talent management processes into intelligent, data-driven ecosystems. Recent advances in Natural Language Processing, transformer architectures, Large Language Models (LLMs), and Retrieval-Augmented Generation (RAG) have enabled significant improvements in… Read full abstract & cite →

Human resource management artificial intelligence generative AI large language models recruitment semantic matching retrieval-augmented generation agentic AI explainable AI systematic literature review

Abstract doi.org/10.14569/IJACSA.2026.0171008 PDF

9

Autonomous Penetration Testing Framework Based on Reinforcement Learning and Intrinsic Motivation

Author 1: Rasha A. Alsakran Author 2: Sultan S. Alqahtani

Ethical penetration testing (PT) is a structured security assessment in which cybersecurity experts simulate real attacks on an organization’s digital assets in a controlled testing environment. The primary objective is to evaluate the effectiveness of cybersecurity controls by detecting vulnerabilities in systems, networks, or applications. Subsequently, report these vulnerabilities to… Read full abstract & cite →

Framework autonomous penetration testing reinforcement learning agent intrinsic motivation

Abstract doi.org/10.14569/IJACSA.2026.0171009 PDF

10

Email Phishing Detection Using NLP and Machine Learning

Author 1: Norah Aljurbua Author 2: Asma A. Alhashmi Author 3: Abdulbasit A. Darem

Phishing emails imitate legitimate correspondence and evolve to evade rule-based filters, yet many reported detection gains rest on single data splits, untested differences, and unexamined benchmark artifacts. This study compares TF-IDF classifiers (Logistic Regression, Decision Tree, SGD, and XGBoost) with recurrent networks (Simple RNN, GRU, Bi-LSTM, and Bi-LSTM with attention)… Read full abstract & cite →

Phishing detection natural language processing TF-IDF recurrent neural networks bidirectional LSTM attention mechanism statistical significance testing dataset provenance email security

Abstract doi.org/10.14569/IJACSA.2026.0171010 PDF

11

An Explainable Machine Learning Framework for AWS S3 CloudTrail-Like Log Analysis in Digital Forensics

Author 1: Aseel Aljuhani Author 2: Asma A. Alhashmi Author 3: Abdulbasit A. Darem

The advent of cloud computing for data storage and management has led to the emergence of advanced cyber threats that call for the need for effective cloud log analysis for digital forensic investigations. While there has been considerable development of machine learning algorithms for anomaly detection, most of the current… Read full abstract & cite →

Cloud forensics cloud audit logs AWS CloudTrail Explainable AI anomaly detection SHAP

Abstract doi.org/10.14569/IJACSA.2026.0171011 PDF

12

Older Adults’ Design Preferences for Augmented Reality in Cultural Tourism: An Experimental Study

Author 1: Nur Amira Suhada Suhaimi Author 2: Zaidatol Haslinda Abdullah Sani

Augmented Reality (AR) has increasing potential to support cultural heritage experiences by presenting historical and cultural information through interactive and immersive media. However, its effectiveness depends on design choices that accommodate users’ preferences and perceptual needs. This study investigated AR design preferences among 11 older adults through seven design preference… Read full abstract & cite →

Augmented reality mid-fidelity prototype cultural heritage older adults design preferences human-computer interaction

Abstract doi.org/10.14569/IJACSA.2026.0171012 PDF

13

Prioritizing Data Quality Improvement Areas in Master Data Management Using the Analytic Hierarchy Process

Author 1: Mohammad Milzam Kasyfillah Hutomo Author 2: Yova Ruldeviyani

Master Data Management (MDM) requires effective data quality management to ensure that master data supports organizational operations and decision-making. However, organizations may face difficulties in determining which data quality improvement areas should receive priority because multiple factors influence data quality. This study aims to prioritize data quality improvement areas in… Read full abstract & cite →

Master data management data quality analytic hierarchy process data governance decision-making

Abstract doi.org/10.14569/IJACSA.2026.0171013 PDF

14

Predicting Item Difficulty and Learner Performance on a Bilingual Gamified Quiz Platform

Author 1: Aibolat Kulatay Author 2: Dinara Zhaisanova Author 3: Bakhytzhan Kulambayev Author 4: Saken Mambetov

Gamified quiz platforms need estimates of question difficulty before response data exist, and learner ratings that support personalized practice. Three related analyses were carried out on separate data sets from a Kazakh–Russian quiz platform. First, lexical and transformer models predicted the observed difficulty of 30,444 items, derived from 177.5 million… Read full abstract & cite →

Educational data mining item difficulty estimation item response theory transformer models Elo rating Kazakh language gamified learning

Abstract doi.org/10.14569/IJACSA.2026.0171014 PDF

15

Static INT8 Quantization of a Conjunctival Anaemia Classifier on Android

Author 1: Enrique Lee Huamani Author 2: Erasmo Montufar-Barrientos Author 3: Victor Romero-Alva Author 4: Vania Yulisa Abanto Castro

This study examines the engineering feasibility and evidential limits of static INT8 post-training quantization for a conjunctival-image anaemia classifier on a mid-range Android phone. Using 202 publicly available subject images, we audited a subject-disjoint partition and trained ResNet-50 on automatically extracted conjunctival regions of interest. On a single held-out set… Read full abstract & cite →

Post-training quantization anaemia conjunctival images Android inference latency patient-level evaluation deployment limitations

Abstract doi.org/10.14569/IJACSA.2026.0171015 PDF

16

A Hybrid Metaheuristic Feature Selection Framework for Enhanced Phishing Detection

Author 1: Abdallah S. Hyassat Author 2: Sumaya Al-Khatib Author 3: Motasem M. Elshourbagy Author 4: Mosleh M. Abualhaj Author 5: Qusai Y. Shambour Author 6: Mohammad Sh. Daoud Author 7: Mohammad A. Alsharaiah Author 8: Mohammed Anbar

Phishing websites remain a critical cybersecurity threat due to their ability to mimic legitimate platforms and deceive users into revealing sensitive information. This study proposes a phishing detection framework that centers on a novel union-based feature selection strategy, WOA union DA, which combines the Whale Optimization Algorithm (WOA) and Dragonfly… Read full abstract & cite →

Phishing machine learning feature selection whale optimization algorithm dragonfly algorithm

Abstract doi.org/10.14569/IJACSA.2026.0171016 PDF

17

Multi-Strategy MO-K-Means Method for Cluster-Center Optimization in Color Image Segmentation

Author 1: Haoqin Chen Author 2: Weibin Kong Author 3: Zhongqing Fang Author 4: Xiaotian Chen Author 5: Xin Shen Author 6: Tiantian Song

Color image segmentation is a fundamental problem in image processing and computer vision. K-means is widely used for clustering-based color image segmentation, but its results are sensitive to the initial cluster centers. In addition, clustering based only on RGB features does not directly use the spatial relationships between pixels. To… Read full abstract & cite →

Color image segmentation K-means clustering Musk Ox Optimizer cluster-center optimization SLIC superpixels

Abstract doi.org/10.14569/IJACSA.2026.0171017 PDF

18

Cryptanalysis and Hardening of Paillier-Based Quantized Consensus Against Weight Inference, Replay, and Ciphertext Manipulation

Author 1: Haewon Byeon

Privacy-preserving consensus protocols are increasingly used to protect distributed control and coordination data, but confidentiality alone does not guarantee that a consensus exchange remains private or correct under repeated observations and active network interference. This study analyzes the 2025 Paillier-based dynamic quantized consensus protocol of Song, Guo, and Wang for… Read full abstract & cite →

Distributed consensus Paillier cryptosystem quantized consensus privacy-preserving control replay attack homomorphic encryption safe POC

Abstract doi.org/10.14569/IJACSA.2026.0171018 PDF

19

Sequence-Based Insider Threat Detection: An Evaluation of LSTM Autoencoders on the CERT Dataset

Author 1: Shouq Alrowili Author 2: Asma A. Alhashmi Author 3: Abdulbasit Abdullah Darem

Insider activity is hard to evaluate through sporadic events since the same actions may occur normally. In this study, we explore the hypothesis that a small history of user activities might be used as a baseline. We work with CERT r4.2 and merge logon, device, HTTP, file, and e-mail logs… Read full abstract & cite →

Insider threat detection User Behavior Analytics (UBA) Long Short-Term Memory (LSTM) autoencoder anomaly detection CERT dataset sequence modeling cybersecurity

Abstract doi.org/10.14569/IJACSA.2026.0171019 PDF

20

Computational Frontiers in Volatility Forecasting: A Systematic Review of Deep Learning Approaches in Financial Markets

Author 1: Mariam Ait Al Author 2: Said Achchab Author 3: Younes Lahrichi

Deep learning is now routinely applied to financial volatility forecasting, but the evidence is dispersed and has not been mapped in a way that separates what the field has established from what it has merely asserted. This study reports a PRISMA-guided systematic review of 186 peer-reviewed studies published between 2002… Read full abstract & cite →

Volatility forecasting deep learning realized volatility systematic literature review graph neural networks financial time series

Abstract doi.org/10.14569/IJACSA.2026.0171020 PDF

21

Beyond Conventional ASR Metrics: Evaluating Semantic Preservation in Low-Resource Speech Recognition

Author 1: Nurul Emilia Diyana Abdul Malik Author 2: Mohd Izzuddin Mohd Tamrin Author 3: Airul Azizan Zainudin Author 4: Muhamad Ariff Ibrahim Author 5: Mohd Zulfaezal Che Azemin

Word error rate (WER) and character error rate (CER) quantify lexical discrepancies in automatic speech recognition (ASR) but do not directly indicate whether the intended message remains intact. This exploratory study compared lexical accuracy with human-rated preservation of essential meaning in Malay-English code-switched conversational speech. Four complete inference pipelines were… Read full abstract & cite →

Automatic speech recognition Malay-English code-switching low-resource speech word error rate character error rate semantic preservation human evaluation exploratory benchmarking

Abstract doi.org/10.14569/IJACSA.2026.0171021 PDF

22

Security Analysis and Hardened VRF-Reputation PBFT for Distributed Medical Diagnostic Consensus

Author 1: Haewon Byeon

Distributed intelligent medical diagnostic systems require consensus mechanisms that remain efficient while preventing malicious or low-quality diagnostic agents from shaping the ledgered clinical decision process. A recent VS-PBFT algorithm combines a verifiable random function (VRF) with dynamic reputation to improve practical Byzantine fault tolerance in medical diagnostic collaboration. This study… Read full abstract & cite →

PBFT verifiable random function reputation mechanism blockchain consensus healthcare security safe POC Byzantine fault tolerance

Abstract doi.org/10.14569/IJACSA.2026.0171022 PDF

23

Ontology-Based Prompt Engineering: A Narrative Review

Author 1: Jin Ma Author 2: Ehab Seedahmed Author 3: Essa Alyounis Author 4: Mohammad Nazir Ahmad

Large Language Models (LLMs) have significantly advanced Prompt Engineering (PE); however, current prompting strategies still rely heavily on unstructured natural language and lack explicit semantic guidance. Ontology has recently emerged as a promising approach for enhancing PE by providing structured domain knowledge and reasoning capabilities, yet existing studies remain fragmented… Read full abstract & cite →

Large language models ontology prompt engineering

Abstract doi.org/10.14569/IJACSA.2026.0171023 PDF

24

LLM-Assisted Offline Reward Design for Reinforcement Learning-Based Irrigation Scheduling

Author 1: Qiufen Wang Author 2: Tianhao Wang Author 3: Ziyue Wang

This study develops an LLM-assisted offline workflow for reward design in reinforcement learning-based irrigation scheduling under a seasonal water allocation. Researchers provide agronomic knowledge and feedback on candidate-policy behavior, and the language model translates these inputs into executable rewards. Eight candidates are compared over four generations. The selected reward, aridity–soil… Read full abstract & cite →

Smart agriculture irrigation scheduling deep reinforcement learning large language models reward design curriculum learning water use efficiency

Abstract doi.org/10.14569/IJACSA.2026.0171024 PDF

25

A Wearable Deep-Learning System for Real-Time Object Detection and Face Recognition to Assist Visually Impaired Navigation

Author 1: Yahia Said Author 2: Mouna Afif Author 3: Abdelnasser Saber Mohamed

The integration of Artificial Intelligence (AI) into wearable assistive technologies offers transformative potential for enhancing mobility, spatial awareness, and autonomy among Blind and Visually Impaired (BVI) individuals. This study introduces a real-time, AI-powered wearable system combining object detection, scene understanding, and facial recognition for safe navigation and environmental interaction. The… Read full abstract & cite →

Assistive mobility blind and visually impaired wearable devices AI in accessibility object detection smart glasses smart cane YOLOv8 optimization scene understanding human-centered navigation

Abstract doi.org/10.14569/IJACSA.2026.0171025 PDF

26

An Edge-AI-Enabled YOLOv26 Framework for Accurate Open-Pit Mine Detection from Aerial Imagery Using NVIDIA Jetson Nano

Author 1: Yahia Said Author 2: Taoufik Saidani Author 3: Mohamed Ben Ammar Author 4: Abdelnasser Saber Mohamed Author 5: Faisal Hadi Alanazi Author 6: Refka Ghodhbani

Open-pit mining sites have to be monitored for environmental management, regulatory compliance, and asset tracking, yet reviewing aerial images of large pits, excavation zones, and extraction areas by hand takes a great deal of time and effort. We propose a real-time framework that runs on edge hardware and detects open-pit… Read full abstract & cite →

Open-pit mine detection aerial imagery object detection YOLO edge AI NVIDIA Jetson Nano

Abstract doi.org/10.14569/IJACSA.2026.0171026 PDF

27

A Multi-Stage Deep Learning Framework for Real-Time Seatbelt Detection in Traffic Surveillance Videos

Author 1: Salah Alghyaline

Seatbelt non-compliance is a major road safety concern. Automated road traffic monitoring using surveillance cameras can support large-scale detection. This study builds a multi-stage deep learning framework for real-time seatbelt detection in roadside traffic surveillance videos. YOLOv8s detects vehicle windshields and defines the region of interest; then it detects occupants… Read full abstract & cite →

Deep learning traffic surveillance seatbelt detection YOLO

Abstract doi.org/10.14569/IJACSA.2026.0171027 PDF

28

Protocol Constraints on Post-Quantum Migration of Offline Authentication in Android HCE Payment System

Author 1: Chaima EHBALI Author 2: Ali Ouacha Author 3: Ahmed El-Yahyaoui

Mobile payment systems built on Host Card Emulation (HCE) inherit cryptographic foundations, such as RSA-based offline data authentication, that were designed long before quantum attacks were a concern. NIST has now finalized its first post-quantum cryptography (PQC) standards, and payment operators are starting to plan migrations, yet little work has… Read full abstract & cite →

Mobile payment security post-quantum cryptography NFC protocol constraints

Abstract doi.org/10.14569/IJACSA.2026.0171028 PDF

29

Exploring Human-Computer Interaction Through Behaviour Analysis Using Artificial Intelligence with Metaheuristic Strategies for Enhanced User Experience

Author 1: Zenah Mahmoud AlKubaisy

The analysis of human-computer interaction (HCI) for mobile devices is a comparatively new research area, which is financially effective and is accessible for a year. Developments in mobile technologies have influenced significant changes in how and when individuals access and utilize data, and are having a penetrating influence on how… Read full abstract & cite →

Human-computer interaction user behavior analysis mobile applications Artificial Intelligence metaheuristic strategies

Abstract doi.org/10.14569/IJACSA.2026.0171029 PDF

30

Dynamic Knowledge Graph Learning Neural Network for Prerequisite-Constrained Personalized Learning Trajectory Recommendation

Author 1: Gulchekhra Abishova Author 2: Nurgali Zaurbekov Author 3: Gulzhigit Toktoshov

Personalized learning trajectory recommendation requires continuous learner knowledge assessment, prerequisite compliance, and adaptive resource sequencing. This study proposes the Temporal Prerequisite-Gated Graph Network (TPG-Net), a dynamic heterogeneous graph learning architecture for generating personalized learning trajectories under pedagogical and time constraints. The proposed methodology integrates educational knowledge graph construction, temporal representation… Read full abstract & cite →

Dynamic knowledge graph personalized learning trajectory prerequisite constraint graph neural network knowledge tracing educational recommendation

Abstract doi.org/10.14569/IJACSA.2026.0171030 PDF

31

Sentiment Classification of Consumer Reviews of Amazon Products Under Severe Class Imbalance: A Reproducible Comparison of Convolutional Neural Networks and Sparse Linear Baselines

Author 1: Shamik Palit Author 2: Preetha V. K. Author 3: Dhanya Sudarsan Author 4: Annie Edwis Threse F. Author 5: Jimy Author 6: Santhuana Sedhumadhavan

Sentiment models that determine the sentiment of online reviews, i.e., programs that can automatically distinguish whether a user’s review is praising or criticizing a product, often need to process data with an extremely large gap between positive and negative samples. The positive and negative samples mentioned here are pre-labeled positive… Read full abstract & cite →

Sentiment analysis amazon products consumer reviews convolutional neural network support vector machine class imbalance reproducibility customer analytics

Abstract doi.org/10.14569/IJACSA.2026.0171031 PDF

32

Recall-Constrained Ensemble Optimization for Imbalanced Financial Distress Prediction: Evidence and Boundary Conditions from Indonesian MSMEs

Author 1: Supriyono Author 2: Purwanto Author 3: Aris Sugiharto

Financial distress prediction for micro, small, and medium enterprises (MSMEs) is a severely imbalanced classification problem in which a missed distress case is far costlier than a false alarm. The prevailing response is a recall-first design that maximizes sensitivity, but such designs yield low precision, so that a large share… Read full abstract & cite →

Financial distress prediction imbalanced classification precision–recall trade-off constrained optimization ensemble learning boundary conditions MSMEs

Abstract doi.org/10.14569/IJACSA.2026.0171032 PDF

33

Agentic Multimodal Violence Detection and Human-Gated Emergency Dispatch

Author 1: ASMAE BAALA Author 2: Mostafa Hanoune Author 3: Mohssine Bentaib

Smart cities deploy dense camera networks, yet automated video analysis usually stops at a detection label, leaving the choice of responder to an overloaded human operator. This study presents an end-to-end multi-agent architecture that carries a violence label from ingestion to an audited, operator-confirmed dispatch decision. Two analyst agents (built… Read full abstract & cite →

Multi-agent systems vision-language models violence detection video surveillance smart cities emergency dispatch human-in-the-loop explainable artificial intelligence

Abstract doi.org/10.14569/IJACSA.2026.0171033 PDF

34

When Retrieval Can Hurt: A Multi-Role 2×2 Human Evaluation of Retrieval and Self-Refinement for Arabic Multiple-Choice Question Generation

Author 1: Abderrahman AZI Author 2: Abdessamad JARRAR Author 3: Moulay AMZIL Author 4: Ahmed EL GHAZI

Retrieval-Augmented Generation (RAG) and iterative self-refinement are often added to large language model (LLM) generation pipelines, but their contribution to the quality of Arabic educational items remains uncertain. This study evaluates Modern Standard Arabic multiple-choice question (MCQ) generation in a road-safety domain with a complete 2×2 design crossing retrieval and… Read full abstract & cite →

Arabic NLP multiple-choice question generation retrieval-augmented generation self-refine human evaluation

Abstract doi.org/10.14569/IJACSA.2026.0171034 PDF

35

A Lightweight Authentication Framework for Scalable Blockchain-Enabled IoT Supply Chain Systems

Author 1: Umi Nabilah Badrol Said Author 2: Mohd Rizuan Baharon Author 3: Hairol Nizam Mohd Shah Author 4: Áine MacDermott

The rapid growth of IoT technologies and blockchain has increased attention to IoT–blockchain integration in various environments. A blockchain is a distributed ledger that records encrypted blocks and stores transactions or digital events shared among entities, which can be verified in the future. Blockchain integration is expected to improve supply… Read full abstract & cite →

Authentication blockchain Internet of Things (IoT) lightweight security supply chain data privacy

Abstract doi.org/10.14569/IJACSA.2026.0171035 PDF

36

An Enhanced DenseNet-169 Framework for Automated Multi-Class Diabetic Retinopathy Classification Using Transfer Learning and SMOTE

Author 1: Sinduja A Author 2: Agila Harshini Thangavel Author 3: Deepa V Author 4: Hemavathi Selvam Author 5: A P Narmadha Author 6: K Abinaya

Diabetic Retinopathy (DR) is the most common microvascular consequence of diabetes mellitus and the most common cause of permanent blindness in the working-age population worldwide. Early diagnosis and quick intervention are critical for preventing severe visual impairment, but traditional screening methods rely on skilled ophthalmologists to manually examine retinal fundus… Read full abstract & cite →

Diabetic retinopathy deep learning DenseNet-169 transfer learning SMOTE convolutional neural networks medical image analysis image classification computer-aided diagnosis

Abstract doi.org/10.14569/IJACSA.2026.0171036 PDF

37

Demand-Driven Competency Extraction from Online Job Postings Using Precision-Constrained Context-Aware Gated Fusion

Author 1: Shi Fengyuan Author 2: Megat F. ZUHAIRI Author 3: Huang Yonghua Author 4: Aabid A. MIR Author 5: Ali TUFAIL Author 6: MH Zafar

Online job postings are rich sources of competency data, yet extraction is challenging because explicit terms, semantic paraphrases, and contextual cues overlap. To address this, we introduce a demand-driven extraction framework centered on Precision-Constrained Context-Aware Gated Fusion (PC-CAGF) to balance recall and precision. We collected 14,030 Chinese IT job postings… Read full abstract & cite →

Competency extraction online job postings natural language processing semantic similarity context-aware gated fusion topic modelling multilabel classification PC-CAGF

Abstract doi.org/10.14569/IJACSA.2026.0171037 PDF

38

Arabic Fake News Detection: A PRISMA-Based Systematic Review of Methods, Datasets, and Challenges

Author 1: Mohammed Fikery Ghemas Author 2: Hesham Ahmed Hassan Author 3: Wafaa Samy El-Kassas

Despite rapid progress in Arabic fake news detection, reported advances remain difficult to interpret because existing studies vary substantially in datasets, language coverage, model architectures, evaluation protocols, and explainability practices. To address this fragmentation, this study conducts a PRISMA-based systematic review of 75 peer-reviewed studies published between 2019 and April… Read full abstract & cite →

Systematic literature review Arabic fake news detection Arabic NLP transformer models hybrid models text classification Arabic datasets

Abstract doi.org/10.14569/IJACSA.2026.0171038 PDF

39

Explainable Aspect-Based Sentiment Analysis for Marine Tourism Using Large-Scale Language Models

Author 1: Aisah Rini Susanti Author 2: Hadiyanto Author 3: Ridwan Sanjaya

Marine tourism is a strategic sector heavily influenced by tourists' perceptions and experiences, which are captured in user-generated content across various digital platforms. However, the multifaceted, informal, and ambiguous nature of reviews poses challenges for sentiment analysis based on conventional methods. This study proposes an aspect-based sentiment analysis (ABSA) framework… Read full abstract & cite →

LLM aspect-based sentiment analysis XAI marine tourism UGC

Abstract doi.org/10.14569/IJACSA.2026.0171039 PDF

40

Cross-Level Scene-Agent Graph Learning for Context-Aware Multi-Label Emotion Recognition

Author 1: K. Suresh Author 2: L. Sherly Pushpa Annabel Author 3: T. Venketbabu Author 4: S. Hariharasudhan Author 5: M.S.S. Sasikumar Author 6: A. Ramar

In context-aware emotion recognition from images, it is necessary to consider the visual appearance of the target person along with the surrounding environment. In unconstrained settings, emotions may be simultaneously related to the target person, thus making the task multi-label. However, existing methods focus on either the representation of the… Read full abstract & cite →

Context-aware emotion recognition multi-label emotion recognition scene-agent graph graph learning cross-level interaction adaptive fusion affective computing EMOTIC

Abstract doi.org/10.14569/IJACSA.2026.0171040 PDF

41

Towards Real-Time Cognitive State Monitoring Using an Explainable Data-Driven Deep Learning Framework on Physiological Signals

Author 1: Mohammed Ibrahim Altwijri

Cognitive states represent dynamic mental conditions that impact how individuals identify their environment, process information, and make decisions. These states are influenced by factors like stress levels, cognitive workload, and external stimuli, which can drastically affect task performance and safety. Therefore, precise examination of cognitive workload is critical in applications… Read full abstract & cite →

Cognitive state monitoring physiological signals feature selection deep learning explainable artificial intelligence real-time systems

Abstract doi.org/10.14569/IJACSA.2026.0171041 PDF

42

Development of a Web-Based Waste Bank Information System for Community Life Insurance Management

Author 1: Paweena Preechayakul Author 2: Chulalak Changul Author 3: Nopporn Bukwan

Municipal solid waste in Thailand accumulates at alarming rates, with household waste contributing over 38.2% of national waste generation. To resolve severe environmental degradation in suburban communities, Wat Khonon Mahasawat established a Community Enterprise operating a waste bank integrated with a mutual life insurance fund. Members deposit segregated recyclables to… Read full abstract & cite →

Waste bank information system community life insurance circular economy technology adoption frameworks CodeIgniter MVC

Abstract doi.org/10.14569/IJACSA.2026.0171042 PDF

43

WHUM: A Hybrid Wavelet Transform-Homomorphic Filtering-Unsharp Masking Method for Chest X-Ray Enhancement

Author 1: Diana Diana Author 2: Ermatita Ermatita Author 3: Abdiansah

The quality of chest X-ray (CXR) images can be greatly impacted by noise, uneven illumination, and low contrast, potentially obscuring anatomical details associated with identifying lung diseases. In this study, we propose Wavelet Homomorphic Unsharp Masking (WHUM), a sequential image-enhancement approach that uses a combination of Wavelet Transform (WT), Homomorphic… Read full abstract & cite →

Wavelet transform homomorphic filtering unsharp masking chest x-ray image enhancement

Abstract doi.org/10.14569/IJACSA.2026.0171043 PDF

44

Predicting Apartment Prices from Arabic Online Listings in Saudi Arabia: Gradient Boosting with Structural, Locational and Text-Derived Features

Author 1: Abdullah M. Baqasah

Reliable price estimation supports housing policy and market transparency in Saudi Arabia, yet the online listings that form the main public source of price data are noisy, heavily duplicated, and describe many attributes only in Arabic free text. This study develops and evaluates a reproducible pipeline for predicting apartment prices… Read full abstract & cite →

House price prediction Saudi Arabia Arabic text features LightGBM XGBoost random forest SHAP data quality real estate listings

Abstract doi.org/10.14569/IJACSA.2026.0171044 PDF

45

Security, Privacy Protection, and Trust in Learning Management System Continuance: Evidence from Sri Lankan Higher Education

Author 1: Nivashinie Gopalaakrishnan Author 2: Anuradha Jayakody

Learning Management Systems (LMSs) are integral to higher education, yet understanding the factors associated with users' intentions to continue using these platforms remains important. While previous continuance research has emphasized functional and experiential factors, the combined relationships among perceived security, perceived privacy protection, trust, and LMS continuance intention warrant further… Read full abstract & cite →

Learning management systems continuance intention perceived security perceived privacy protection institutional platform trust higher education Sri Lanka

Abstract doi.org/10.14569/IJACSA.2026.0171045 PDF

46

Actionable and Trustworthy AI for Student Depression Prediction Using Multi-Level Explainability

Author 1: Sanjit Kumar Dash Author 2: Deepak Kumar Pradhan Author 3: Mohammed Altaf Ahmed Author 4: Qutubuddin Mohammed

It is critical to be able to identify and intervene with high-risk students as early as possible - this is most easily accomplished by developing scalable systems for providing such early warning signals. Although many researchers have developed highly accurate machine learning (ML) models to detect depression, poor clinical transparency… Read full abstract & cite →

Explainable AI (XAI) student depression stacking ensemble data security SMOTE SHAP LIME Counterfactual Explanations (DiCE)

Abstract doi.org/10.14569/IJACSA.2026.0171046 PDF

47

Non-Functional Requirement (NFR) Characteristics for Cloud-Based IoT Systems: A Systematic Literature Review

Author 1: Nor Faradila Kolan Author 2: Fazlina Mohd Ali Author 3: Elankovan A. Sundararajan Author 4: Marizuana Mat Daud Author 5: Siti Nur’asyiqin Ismael

The rapid development and increasing integration of the Internet of Things (IoT) and cloud computing can improve a system's ability to process big data in real time. These technologies complement each other. A cloud-based IoT system is more complex and presents new challenges in requirements engineering, especially for non-functional requirements… Read full abstract & cite →

Non-functional requirement software engineering requirement engineering Internet of Things cloud computing

Abstract doi.org/10.14569/IJACSA.2026.0171047 PDF

48

Adaptive Neural Network PID Controller for Quadcopter UAVs Trajectory Tracking

Author 1: Lafi ALNUFAIE

Quadcopter unmanned aerial vehicles (UAVs) are widely used in civilian and military missions owing to their vertical takeoff and landing capability, agility, and versatility, yet their practical deployment critically depends on accurate trajectory tracking under wind, payload variations, sensor noise, and other uncertainties. Trajectory tracking control remains challenging because quadcopter… Read full abstract & cite →

Quadcopter UAV trajectory tracking adaptive PID control neural network

Abstract doi.org/10.14569/IJACSA.2026.0171048 PDF

49

Semantic Feature Clustering Using LLM Embeddings for Label-Dependent Feature Selection

Author 1: Shikha Choudhary Author 2: Bhawna Saxena

Most feature selection methods assess feature relevance using individual scoring criteria followed by a separate selection step. Although clustering-based approaches may at times capture relationships among features, relevance and redundancy are not always addressed in unison. Consequently, redundant features may receive high relevance scores and get selected, possibly leading to… Read full abstract & cite →

Feature selection semantic clustering large language model embeddings label-dependent dimensionality reduction

Abstract doi.org/10.14569/IJACSA.2026.0171049 PDF

50

Feature Selection for Machine Learning and Deep Learning-Based Heart Disease Prediction: A Systematic Review of Predictive Performance and Evaluation Practices

Author 1: Wan Aezwani Wan Abu Bakar Author 2: Mustafa Man Author 3: Nurnadiah Zamri Author 4: Ily Amalina Ahmad Sabri Author 5: Noor Suhana Sulaiman Author 6: Azliza Yacob Author 7: Nor Sukinah Aziz Author 8: Sharif Ahmad Author 9: Nur Laila Najwa Josdi

The persistent global burden of cardiovascular disease has driven growing interest in machine-learning (ML) and deep-learning (DL) approaches for heart disease prediction. Structured clinical datasets often contain redundant or weakly informative variables, prompting the integration of feature-selection techniques into predictive pipelines. However, whether feature selection improves prediction on its own… Read full abstract & cite →

Deep learning feature selection clinical target taxonomy heart disease prediction information leakage machine learning multilayer perceptron recursive feature elimination systematic review

Abstract doi.org/10.14569/IJACSA.2026.0171050 PDF

51

Geolocation and Gamification to Assess Mathematical Reasoning: Motivational and Innovative Strategy

Author 1: Norka Bedregal-Alpaca Author 2: Sidanelia Flores-Silva Author 3: Víctor Cornejo-Aparicio Author 4: Karim Guevara

The development of logical-mathematical reasoning skills is essential for fostering problem-solving and critical thinking. However, traditional teaching and assessment methods may limit the practical application of these skills. Exam-related tension often causes mental and physiological blocks that affect performance, making it necessary to design motivational assessment strategies that reduce anxiety… Read full abstract & cite →

Evaluation motivation for studies practical examination mathematic game educational innovation

Abstract doi.org/10.14569/IJACSA.2026.0171051 PDF

52

Reducing Energy Shortages and External Dependence in Tunisia: A Renewable Hybrid Energy Framework Through 2040

Author 1: Mounir Bouzguenda

Tunisia’s power system faces increasing exposure to declining domestic hydrocarbon production, imported natural gas, and regional electricity purchases. Official statistics show that primary-energy resources decreased from 4.436 Mtoe in 2023 to 3.378 Mtoe in 2025, while energy independence fell from 49% to 35%. This study updates the national baseline through… Read full abstract & cite →

Battery energy storage capacity adequacy ELMED energy security hybrid energy system renewable integration Tunisia wind capacity factor

Abstract doi.org/10.14569/IJACSA.2026.0171052 PDF

53

Comparative Evaluation of Classical CPU Scheduling Algorithms in a Multicore Simulation Environment

Author 1: Mirjana Kocaleva Vitanova Author 2: Biljana Zlatanovska Author 3: José Alejandro Ramón Rocha Author 4: Aleksandar Krstev Author 5: Elena Karamazova Gelova Author 6: Marija Miteva Author 7: Zoran Zlatev

Classical CPU scheduling algorithms remain useful baselines for understanding the trade-offs among waiting time, turnaround time, responsiveness, processor utilization, and scheduling overhead. This study compares First-Come, First-Served (FCFS), non-preemptive Shortest Job First (SJF), Shortest Remaining Time First (SRTF), and Round-Robin (RR) using a custom discrete-event simulator configured with a global… Read full abstract & cite →

CPU scheduling FCFS SJF SRTF Round-Robin multicore simulation discrete-event simulation performance evaluation

Abstract doi.org/10.14569/IJACSA.2026.0171053 PDF

54

Explaining and Predicting Public Transportation Adoption Among Female University Students in Saudi Arabia: A TAM and Machine Learning Approach

Author 1: Waad Bouaguel Author 2: Chiheb Eddine Ben Ncir

Public transportation is important to sustainable mobility in Saudi Arabia, but adoption among female university students remains underexplored. This study examines behavioral intention using an extended Technology Acceptance Model comprising perceived usefulness, perceived ease of use, perceived safety, privacy and comfort, social influence, service quality, perceived affordability, and environmental concern… Read full abstract & cite →

Public transportation machine learning technology acceptance model TAM PLS-SEM data-driven computational approaches sustainable mobility

Abstract doi.org/10.14569/IJACSA.2026.0171054 PDF

55

Continuous Integration and Continuous Delivery Ready Defect Prediction for Risk Aware Test Case Prioritisation

Author 1: Jijo Joseph Cheeramban George Author 2: D. Peter Augustine

Continuous Integration and Continuous Delivery pipelines are highly dependent on automated regression tests to enable quick and effective feedback to the developers. Test Case Prioritisation provides a useful approach by arranging the tests in a sequence such that the ones that are likely to yield most faults are run first… Read full abstract & cite →

Continuous integration and continuous delivery pipelines defect prediction test case prioritisation risk-aware testing regression testing

Abstract doi.org/10.14569/IJACSA.2026.0171055 PDF

56

Validating Selective Prediction Under Source Uncertainty in Small Enterprise Surveys

Author 1: Suyono Thamrin Author 2: M. Calvin Capnary Author 3: Iskandar Agung

Survey-based prediction is sensitive to unresolved source versions, repeated predictor profiles, and abstention rules. We evaluate these choices jointly in two Indonesian culinary-enterprise files, containing 206 aligned positions and 301 recorded rows, respectively. Each file has 177 predictor profiles, without verified respondent identities. Five fixed baselines are compared over 50… Read full abstract & cite →

Selective prediction conformal prediction data provenance grouped validation survey data small enterprises

Abstract doi.org/10.14569/IJACSA.2026.0171056 PDF

57

Deformable Attention-Guided Multi-Modal Transformer Framework for Accurate Brain Tumor Subregion Segmentation

Author 1: Guldana Shaimerdenova Author 2: Aigerim Toktarova Author 3: Zhanna Zhunisbekova Author 4: Elvira Adylbekova Author 5: Gulmira Baenova Author 6: Aiman Abibullayeva

Accurate segmentation of brain tumor subregions from multi-modal magnetic resonance imaging remains a challenging task due to heterogeneous tumor appearance, ambiguous boundaries, and complex spatial structures. This study proposes a deformable attention-guided multi-modal transformer framework designed to address these challenges through efficient context modeling and adaptive feature representation. The architecture… Read full abstract & cite →

Brain tumor multi–modal learning deformable attention intersection over union deep learning

Abstract doi.org/10.14569/IJACSA.2026.0171057 PDF

58

Detection of Necrosis and Steatosis in Liver Histopathology Images Using a CARAFE- and Wise-IoU-Enhanced YOLOv8

Author 1: Lailil Muflikhah Author 2: Putu Wahyu Kusuma Wardhana

Automated localization of necrosis and steatosis in liver histopathology images is challenging because tissue regions exhibit heterogeneous appearances and indistinct boundaries. This preliminary study evaluates the integration of two established components, Content-Aware ReAssembly of FEatures (CARAFE) and Wise-IoU v3, into YOLOv8s for detecting necrosis, steatosis, and normal cellular regions. CARAFE… Read full abstract & cite →

Liver histopathology hepatic necrosis hepatic steatosis object detection YOLOv8 CARAFE Wise-IoU transfer learning

Abstract doi.org/10.14569/IJACSA.2026.0171058 PDF

59

Extending HSV-Based Color Selection for Non-Invasive Plant Disease Detection Using UAV Visible-Light Imagery

Author 1: Toni Kusnandar Author 2: Heru Nugroho Author 3: Erna Hikmawati Author 4: Aisah Nur Endah Sari Author 5: Fitri Rizqiati

Per-pixel HSV color selection methods from prior UAV-based research do not generalize across imaging sessions. This study evaluates plot-level aggregated RGB and HSV statistics of vegetation pixels (separated using the Excess Green index with Otsu thresholding) with a Random Forest, on 732 UAV RGB plot images from a multi-year common… Read full abstract & cite →

UAV HSV color selection excess green random forest radiometric calibration common bean

Abstract doi.org/10.14569/IJACSA.2026.0171059 PDF

60

Multimodal Cardiac Phenotype Estimation Using ECG and Clinical Data

Author 1: Kavitha Bai A. S. Author 2: J. Somasekar

Existing ECG-based cardiac assessment methods primarily focus on disease classification, providing limited information about quantitative cardiac function, structural remodeling, and hemodynamic status. The complex relationship between ECG patterns and echocardiographic phenotypes remains insufficiently explored. This study proposes a Multimodal Cardiac Phenotype Prediction Network (MultiCard-PNet) for estimating clinically relevant cardiac phenotypes… Read full abstract & cite →

Cardiac function phenotype estimation electrocardiogram 1D-ResNet multi-layer perceptron

Abstract doi.org/10.14569/IJACSA.2026.0171060 PDF

61

Deep Learning-Based Touch-Event Recognition from Individual RGB Frames: A Comparative Study of CNN Architectures and Model Training Strategies

Author 1: Arash Ghazvini Author 2: Irwandi Hipiny Author 3: Mazdan Ali Amaran Author 4: Hasnizam Abdul Wahid Author 5: Johari Abdullah Author 6: Hamimah Ujir

Touch-event recognition from visual data remains a challenging computer vision task, particularly when distinguishing the body regions involved in physical interactions. Unlike video-based action recognition, frame-level touch-event recognition requires models to identify interaction classes from individual RGB images without relying on temporal motion information. However, the effectiveness of different convolutional… Read full abstract & cite →

Touch-event recognition human interaction recognition deep learning convolutional neural networks RGB frames class-imbalanced learning

Abstract doi.org/10.14569/IJACSA.2026.0171061 PDF

62

Hybrid ANOVA and DOA-GA Feature Selection for EEG-Based Emotion Recognition

Author 1: Ghassan G. Kanaan Author 2: Mu'taz Al-Zghoul Author 3: Tarek Kanaan

Using a Dhole Optimization Algorithm-Genetic Algorithm (DOA-GA) wrapper, this study reduces feature numbers from an ANOVA filter retaining 500 features. DOA and GA utilize half of the iteration budget, and the fitness function balances subset size and error using K-nearest neighbors. EEG Brainwave and HBN datasets were 4-channel and 128-channel… Read full abstract & cite →

Electroencephalography emotion recognition feature selection metaheuristic optimization dhole optimization algorithm genetic algorithm

Abstract doi.org/10.14569/IJACSA.2026.0171062 PDF

63

Modeling of Artificial Intelligence-Driven Context-Aware Memotion Analysis Using Integrated Large-Scale Multimodal Corpora

Author 1: Rania Aboalela

Memotion analysis is a complex task designed to accurately identify the sentiments and feelings conveyed within internet memes. A meme is any trend, skill, or belief learned from others, rather than inherited genetically, and shared through imitation. The spread of malicious and aggressive memes has aggravated issues on social media… Read full abstract & cite →

Memotion large-scale multimodal data artificial intelligence feature extraction RoBERTa-large embeddings model optimization

Abstract doi.org/10.14569/IJACSA.2026.0171063 PDF

64

Nearest-Neighbor Case Retrieval for Emergency Department Urgency Prediction with a Large Language Model

Author 1: Ismail Bouajaja Author 2: Omar Bouattane Author 3: Oussama Barakat Author 4: Abdelaziz Daaif

This study evaluates whether prior emergency department visits improve a large language model’s (LLM) prediction of recorded urgency. Four methods were compared: an always-urgent baseline, conventional k-nearest-neighbor (kNN) voting, an LLM without retrieved cases, and the same LLM with 30 kNN-retrieved labeled cases. The 2022 National Hospital Ambulatory Medical Care… Read full abstract & cite →

Emergency department urgency prediction large language models nearest neighbors case retrieval tabular classification

Abstract doi.org/10.14569/IJACSA.2026.0171064 PDF

65

Accuracy and Reliability of Monocular 3D Pose Estimation for Lower-Limb Movement Analysis

Author 1: Guangjun Wang Author 2: Jiaojiao Chen Author 3: Mengxia Hu Author 4: Cezhong Luo Author 5: Kaibo Li Author 6: Xiaohui Chen

Optical motion capture provides accurate joint kinematics, but its cost and setup complexity limit routine use in rehabilitation and home monitoring. Monocular RGB video is a low-cost alternative, yet many 3D pose estimation studies report overall errors on public benchmarks without distinguishing lower-limb movements or individual joints. This study evaluated… Read full abstract & cite →

3D pose estimation monocular RGB lower-limb analysis accuracy noise robustness temporal error analysis

Abstract doi.org/10.14569/IJACSA.2026.0171065 PDF

66

MIR-YOLO: Morphology-Invariant and Reliability-Aware YOLO for Generalizable Building-Surface Defect Detection

Author 1: Dongmei Zhao Author 2: Yu Gu

Reliable building-surface inspection requires transfer beyond training sites, structures, and acquisition sequences. We evaluate MIR-YOLO, a YOLO11n-derived detector with paired-view appearance diversification and feature/prediction consistency, edge-guided P2–P4 fusion, localization-quality estimation, and source-validation-only isotonic calibration. On the MBDD2025 building-grouped test, mAP50–95 improved from 0.495 to 0.519 over five matched seeds (Δ… Read full abstract & cite →

Building surface defects automated inspection object detection domain generalization confidence calibration aerial imagery YOLO

Abstract doi.org/10.14569/IJACSA.2026.0171066 PDF

67

Time Series Analysis and Forecasting of University Enrolment Using Grid Search–Optimized Holt–Winters Algorithm: Evidence from Tawi-Tawi, Philippines

Author 1: Irene A. Bakili Author 2: Charisa L. Abalie Author 3: Nerico L. Mingoc Author 4: Romeo L. Pañares Jr. Author 5: Jul-habir U. Kali Author 6: Ramat B. Hailaya Author 7: Dayang Farha L. Julambri Author 8: Estelita Cheng G. Lim Author 9: Abdul-aziz J. Tiblan Author 10: Urbano B. Patayon

Student enrolment forecasting supports institutional planning and resource allocation. This study employed grid-based structural selection of Holt–Winters Exponential Smoothing models to analyse enrolment records from Mindanao State University–Tawi-Tawi College of Technology and Oceanography covering academic years 2011–2012 to 2025–2026. A total of 69,103 student records were aggregated by semester, academic… Read full abstract & cite →

Enrolment forecasting Holt-Winters exponential smoothing grid search optimization higher education trends time series analysis

Abstract doi.org/10.14569/IJACSA.2026.0171067 PDF

68

Mortality Risk Prediction in Non-Alcoholic Fatty Liver Disease Through a Modified Stacked Gated Recurrent Unit Using Clinical Data

Author 1: Jayashree Patil Author 2: Shwetambari Chiwhane Author 3: Varsha Lokare

Non-Alcoholic Fatty Liver Disease (NAFLD) is a progressive liver disease caused by excessive fat accumulation in hepatocytes in the absence of significant alcohol intake. Liver biopsy remains the diagnostic reference standard but is invasive and costly, limiting its use. In this study we address a distinct task namely mortality risk… Read full abstract & cite →

Non-alcoholic fatty liver disease modified stacked gated recurrent unit refine gate mechanism NAFLD machine learning

Abstract doi.org/10.14569/IJACSA.2026.0171068 PDF

69

Human-Guided Decision Support for Selective Harvesting in Tropical Forests: Scenario Simulation, Growth Prediction, and Expert Evaluation at Pasoh Forest Reserve

Author 1: Nabila Wardah Zamani Author 2: Nurfadhlina Mohd Sharef Author 3: Mohd Hafeez Osman Author 4: Raihani Mohamed Author 5: Nurul Amelina Nasharuddin Author 6: Razali Yaakob Author 7: Syaifulnizam Abd Manaf Author 8: Warren Marc Carvalho Author 9: Lee Lai Soon

Sustainable forest management requires analytical tools that enable forest managers to evaluate how harvesting decisions affect future forest structure and ecosystem services. An increase in the reliability of predicting how trees will develop has occurred because of the use of models and machine learning techniques over the past several years… Read full abstract & cite →

Decision support system scenario simulation selective harvesting forest growth prediction selective management system human-centred artificial intelligence

Abstract doi.org/10.14569/IJACSA.2026.0171069 PDF

70

Performance Evaluation of a Sentiment-Enhanced LSTM Framework for Portfolio Management Using Sharpe Ratio, Drawdown, and Prediction Accuracy

Author 1: Minal Dhankar Author 2: Neha Gupta

While the price aspects in volatile markets do need to be included in the portfolio management process, existing quantitative models have failed to take into account the behavioural influences. In this study, the performance of a sentiment-enhanced LSTM framework for portfolio management is compared to benchmarks with 10,000 observations of… Read full abstract & cite →

Deep learning LSTM natural language processing sentiment augmentation

Abstract doi.org/10.14569/IJACSA.2026.0171070 PDF

71

Time-Frequency Feature Fusion with Multi-Scale Attention for Underwater Acoustic Target Recognition

Author 1: Gang Hu Author 2: Yifei Song Author 3: Mohamad Farhan Mohamad Mohsin

Traditional convolutional neural networks (CNNs) for underwater acoustic target recognition (UATR) are limited by restricted local receptive fields, and reliance on frequency-domain inputs alone can discard temporal information present in the raw waveform. This study proposes TFMSUATR, a model that fuses a raw-waveform branch and a log-magnitude spectrogram branch through… Read full abstract & cite →

Underwater acoustic target recognition time-frequency feature fusion multi-scale attention Swin Transformer global dependency modeling

Abstract doi.org/10.14569/IJACSA.2026.0171071 PDF

72

Secure Intelligent Edge: A Lightweight Privacy-Preserving Scheme for Edge Computing Using Query-Adaptive Inner Product Functional Encryption

Author 1: Thejaswini S Author 2: R Aparna

Recently, the digitalization era has led to a demand for secure digitalization transformation due to the dataplexity and secureplexity in modern applications; protecting sensitive data against unauthorized exploitation while extracting actionable intelligence has become a critical security necessity. In this context, Functional Encryption (FE) is used for privacy-preserving data analytics… Read full abstract & cite →

Functional encryption inner product functional encryption lightweight modular arithmetic communication- computation efficiency edge computing privacy preserving

Abstract doi.org/10.14569/IJACSA.2026.0171072 PDF

73

LifeChain: A Hybrid Cross-Chain Blockchain Framework for Interoperable and Integrity-Preserving Organ Donation and Transplant Matching

Author 1: Khalida Tabassum Author 2: Syed Imtiyaz Hassan

Organ allocation depends on timely and trustworthy exchange of donor and recipient information among hospitals that run independent information systems. Existing blockchain solutions for healthcare usually rely on a single ledger, which forces a trade-off between the confidentiality required for donor data and the public auditability required for allocation decisions… Read full abstract & cite →

Blockchain interoperability cross-chain communication hyperledger fabric hyperledger cactus ethereum data integrity organ donation

Abstract doi.org/10.14569/IJACSA.2026.0171073 PDF

74

Development of an Edge-AI Autonomous UAV Control System for Navigation and Perception in GPS-Denied Environments

Author 1: Alizhan Tulembayev Author 2: Gani Baiseitov Author 3: Zarina Kutpanova Author 4: Alexey Semchenko Author 5: Dmitriy Shandronov Author 6: Praveen Kumar

In aerospace research on the cross-cutting growth of unmanned aerial vehicles (UAVs) in inspection, monitoring, and emergency response use cases, reliable operation in environments that are constrained by navigation and communication infrastructure is an issue of vital importance. Conventional UAV systems have a strong reliance on Global Positioning System (GPS)… Read full abstract & cite →

Unmanned Aerial Vehicle (UAV) GPS-denied environments visual–inertial sensors embedded computing real-time object recognition

Abstract doi.org/10.14569/IJACSA.2026.0171074 PDF

75

Deployment-Stage Validation and Web-Based Screening of Hybrid Ensemble Models for Workplace Depression and Anxiety Prediction

Author 1: Akib Tanzim Author 2: Ramisha Binte Helal Author 3: Nyme Ahmed

This study assesses the deployment stage model refitting and consistency of implementation for hybrid stacking classifiers to classify the PHQ-9 and GAD-7 severity for 406 office workers from Bangladesh using questionnaires. The original audit compared evaluation models built with 284 training records with archived deployment models built with 345 development… Read full abstract & cite →

Workplace mental health depression-anxiety severity PHQ-9 GAD-7 model retraining hybrid ensemble learning

Abstract doi.org/10.14569/IJACSA.2026.0171075 PDF

76

Quantum Computing for Cardiovascular Imaging: A Review of Emerging Applications, Challenges and Future Perspectives

Author 1: Aisyah Mohd Rahimi Author 2: Amirul Asyraf Mohamad Zhahir Author 3: Azira Khalil Author 4: Muhammad Mokhzaini Azizan

Quantum computing is being investigated as a computational approach for medical image analysis, but its potential role in cardiovascular imaging remains unclear. This review examines gate-based, quantum annealing, and hybrid quantum-classical techniques for cardiovascular image enhancement, image reconstruction, image segmentation, image feature extraction, and image-based disease classification, and it treats… Read full abstract & cite →

Quantum computing cardiovascular imaging quantum machine learning hybrid quantum-classical computing medical image analysis

Abstract doi.org/10.14569/IJACSA.2026.0171076 PDF

77

A Data-Driven Multi-Objective Optimization Framework for Rear-End Crash Compatibility Design of Truck Rear Underrun Protection Devices

Author 1: Yansong Dai Author 2: Xiao Zhao Author 3: Yidi Ma Author 4: Cong Wang Author 5: Hongrui Nie

To mitigate underride risks in passenger car–truck rear-end collisions caused by mismatches in vehicle mass, structural stiffness, and height, this study proposes a surrogate model-assisted multi-objective optimization framework for rear underride protection device design. A full-overlap passenger car–truck crash model was first established using explicit dynamic finite element analysis and… Read full abstract & cite →

Rear underride protection device rear-end crash compatibility explicit dynamic finite element method Kriging surrogate model NSGA-II multi-objective optimization

Abstract doi.org/10.14569/IJACSA.2026.0171077 PDF

78

Adaptive Fuzzy Clustering with Granular Refinement and Triangulation

Author 1: Anis Farhani Abdul Ghafar Author 2: Syarifah Sakinah Syed Ahmad

Traditional fuzzy C-means (FCM) typically uses a fixed Euclidean distance, which may be restrictive for heterogeneous or overlapping data. This working manuscript examines GFASM-FCM, which combines iteration-wise metric selection, ambiguity-focused membership sharpening, and lower-approximation prototype refinement. The originally reported 600-point synthetic results were Silhouette 0.82, DBI 0.23, PC 0.66, CE… Read full abstract & cite →

Adaptive distance metric fuzzy C-means fuzzy clustering granular computing overlap verification triangulation unsupervised learning

Abstract doi.org/10.14569/IJACSA.2026.0171078 PDF

79

Prediction Versus Identification: Machine Learning and Small-Sample Inference for Public Credit Guarantees to Small and Medium-Sized Enterprises in Seven European Countries

Author 1: Abdellah HAFID Author 2: Anouar RIAD SOLH Author 3: Hamid EZ-ZAHRAOUY

Public credit guarantees for small and medium-sized enterprise (SME) lending, long-standing instruments of financing policy, have seen their use intensify across Europe during episodes of credit stress, particularly during the pandemic. Their effect on credit volumes nonetheless remains poorly established at the aggregate level. This article examines whether these schemes… Read full abstract & cite →

SME credit public credit guarantees wild cluster bootstrap resampling-based inference statistical power double machine learning prediction versus identification dynamic panel

Abstract doi.org/10.14569/IJACSA.2026.0171079 PDF

80

Artificial Intelligence-Driven Hybrid Ensemble Model for Cyber Attack Detection in Internet of Things-Enabled Cloud Systems

Author 1: Anwer Mustafa Hilal Author 2: Salahaldeen Duraibi

The Internet of Things (IoT) is an expanding network of interconnected devices, sensors, and applications that share and communicate data via the internet. Sustainable Cities and Communities is a one main goal of Sustainable Development Goal (SDG) which aims to make urban areas inclusive, safe, resilient, and sustainable by 2030… Read full abstract & cite →

Artificial intelligence cyber attack detection ensemble model dimensionality reduction Internet of Things cloud environment

Abstract doi.org/10.14569/IJACSA.2026.0171080 PDF

81

Temporal Evidence-Guided Multiset-Preserving Reallocation for Cross Domain Weakly Supervised Video Anomaly Localization

Author 1: Marwan Tayseer Al Zawahra Author 2: Mohammad Subhi Al-Batah Author 3: Mowafaq Salem Alzboon

Cross-domain weakly supervised video anomaly localization may fail even when a detector produces useful anomaly evidence, because high anomaly scores can be attached to the wrong temporal segments. This failure mode is studied through TEMRAL, a deterministic post-inference procedure for frozen weakly supervised detectors. TEMRAL measures local changes between neighboring… Read full abstract & cite →

Weakly supervised video anomaly detection cross domain generalization temporal anomaly localization multiple instance learning post inference reallocation multiset preservation temporal representation change

Abstract doi.org/10.14569/IJACSA.2026.0171081 PDF

82

Voice Identification Based on the Use of Chirplet Transforms

Author 1: Aliaskar Magzhan Author 2: Mazakov Talgat Author 3: Mohamed Othman Author 4: Jomartova Sholpan Author 5: Mazakova Aigerim

Voice biometrics is based on extracting stable individual features from a speech signal. However, practical identification accuracy is significantly reduced by short fragments, variations in speech rate, noise, and mismatched recording channels. This study explores the potential of chirplet transforms as a specialized tool for analyzing the non-stationary time-frequency structure… Read full abstract & cite →

Voice identification voice biometrics chirplet transform time-frequency analysis non-stationary signal instantaneous frequency speaker embeddings noise resistance

Abstract doi.org/10.14569/IJACSA.2026.0171082 PDF

83

An Integrated IoT-Enabled Smart Stick for Visually Impaired Navigation

Author 1: Mai Samir Mohamed Author 2: Mohamed Kayed Author 3: Al Hussien Seddik Saad Author 4: Marwa Hashem Shehata

Visual impairment is a profound global health challenge that significantly impacts the quality of life, independence, and safety of millions of individuals. This paper presents an integrated IoT-enabled smart stick for visually impaired people to navigate their surroundings, with sensing modes for enclosed and open spaces. The proposed smart stick… Read full abstract & cite →

Assistive technology smart stick visually impaired YOLOv8 Raspberry Pi Time-of-Flight (ToF) distributed computing

Abstract doi.org/10.14569/IJACSA.2026.0171083 PDF

84

GeoRank360-CNN: Spherical Geometry-Aware Visual Quality Assessment and Optimization for 360° Immersive Digital Media Scenes

Author 1: Weiya Yang Author 2: Liang Chen

In 360° immersive digital media, visual quality assessment is crucial for maintaining perceptual fidelity, as compression distortion and equidistant cylindrical projection jointly affect the viewing quality. This study proposes a spherical geometric perception framework named GeoRank360-CNN for all-round image quality assessment and quality-guided optimization. This framework uses ConvNeXt-Tiny as the… Read full abstract & cite →

Omnidirectional image quality assessment immersive digital media spherical geometry-aware convolution quality prior fusion quality-guided optimization

Abstract doi.org/10.14569/IJACSA.2026.0171084 PDF

85

Covariance-Aware Multi-Backbone Feature Fusion with Bidirectional Temporal Modelling for Frame-Level Video Anomaly Detection

Author 1: Umar Shafiu Haruna Author 2: Abdulrauf Garba Sharifai Author 3: Muktar Danlami Author 4: Aminu Abdulkadir Mahmoud Author 5: Lawan Garba

Video anomaly detection remains challenging because abnormal events are rare, diverse, temporally dependent, and strongly influenced by scene context. This study evaluates a modular covariance-aware framework combining five frozen ImageNet-pretrained CNN backbones (VGG19, ResNet50, EfficientNetB0, DenseNet121, and InceptionV3), feature standardisation, principal component analysis (PCA), Mahalanobis normality scoring, early and late… Read full abstract & cite →

Video anomaly detection Mahalanobis distance principal component analysis CNN feature fusion BiTCN temporal modelling UCSD Ped2 avenue ShanghaiTech

Abstract doi.org/10.14569/IJACSA.2026.0171085 PDF

86

A Lightweight Federated-Learning Intrusion Detection System for Agricultural IoT: Random-Forest-Guided Feature Selection with a Temporal Convolutional Classifier

Author 1: Mohd Abdul Rahim Khan Author 2: Yasir Hashim Naif

Internet of Things (IoT) devices used in precision agriculture, such as soil and climate sensors, irrigation controllers, livestock trackers, and autonomous machinery, form a large, heterogeneous, and often poorly secured attack surface, referred to here as Agricultural IoT (Ag-IoT). Centralized intrusion detection is difficult to apply in this setting because… Read full abstract & cite →

Federated learning intrusion detection system agricultural internet of things temporal convolutional network random forest feature selection federated averaging non-independent and identically distributed data edge computing lightweight deep learning

Abstract doi.org/10.14569/IJACSA.2026.0171086 PDF

87

IoT Security Maturity and Resilience in the Saudi Smart Tourism Ecosystem

Author 1: Ali Abu Abid Author 2: Sultan Algarni

As Internet of Things (IoT) technologies become more deeply embedded in smart tourism, connected infrastructure has grown into a core operational dependency, and service continuity now needs to be treated as a complement to conventional security controls rather than an afterthought. This study evaluates IoT security maturity in the Saudi… Read full abstract & cite →

IoT security cyber resilience smart tourism Saudi vision 2030 regulatory alignment availability threat taxonomy

Abstract doi.org/10.14569/IJACSA.2026.0171087 PDF

88

FD-Square-Det: Frequency-Guided Robust Pedestrian Detection Under Adversarial and Cross-Domain Conditions

Author 1: Prithiraj Modak Author 2: Monjurul Islam Nimon Author 3: Tanvir Ahmed Author 4: Rifath Mahmud Author 5: Syeda Anika Tasnim

Pedestrian detectors face physical adversarial patches and distribution shift between training and deployment data. These threats are usually studied separately. This study asks whether robustness to patches and cross-domain generalization can improve together or conflict. Detectors based on You Only Look Once version 8 medium (YOLOv8m) are evaluated under a… Read full abstract & cite →

Adversarial patch cross-domain generalization domain adaptation explainable artificial intelligence frequency-domain learning pedestrian detection YOLOv8

Abstract doi.org/10.14569/IJACSA.2026.0171088 PDF

89

Beyond Intra-Dataset Accuracy: Cross-Dataset Generalization of Scratch-Trained and Pretrained CNNs for Face Presentation Attack Detection

Author 1: Nawal Alsufyani

Face presentation attack detection (PAD) systems often achieve strong performance under intra-dataset evaluation but experience substantial degradation when exposed to unseen acquisition conditions. This study presents an empirical investigation of the discrepancy between intra-dataset performance and cross-dataset generalization by comparing a lightweight convolutional neural network trained from scratch (SimpleCNN) with… Read full abstract & cite →

Face presentation attack detection face anti-spoofing cross-dataset evaluation domain shift transfer learning deep learning biometrics

Abstract doi.org/10.14569/IJACSA.2026.0171089 PDF

90

Hybrid Deep Learning for EEG-Based Intelligence Prediction: Combining PSD, MSE, and CNN-GRU

Author 1: Mohammed Alrowaili Author 2: Mohammed J. Alhaddad Author 3: Mutasem Jarrah Author 4: Saim Rasheed

Electroencephalography (EEG) signals provide a non-invasive means of examining patterns of brain activity associated with cognitive processing. This study investigates whether spectral and complexity-based EEG characteristics can support the classification of intelligence quotient (IQ) levels using a hybrid deep-learning framework. EEG data were obtained from a publicly available dataset comprising… Read full abstract & cite →

Electroencephalogram (EEG) intelligence prediction Power Spectral Density (PSD) Multiscale Entropy (MSE) deep Learning CNN–GRU Brain–Computer Interface (BCI) signal processing

Abstract doi.org/10.14569/IJACSA.2026.0171090 PDF

91

HCP-RDD: Hierarchical Caption-Prefix Framework for Road Damage Detection Based on YOLO11s

Author 1: Shenghua Liu Author 2: Tianci Ao Author 3: Peng Shao

Road damage detection identifies pavement distresses such as cracks using vision technologies to support road safety. Most existing methods rely solely on visual features and provide poor fine-grained recognition under illumination variation and shadows. Unconstrained noisy scene captions also degrade visual feature learning. This study proposes a Hierarchical Caption-Prefix framework… Read full abstract & cite →

Road damage detection YOLO11s caption conditioning cross-attention multi-scale fusion ablation study

Abstract doi.org/10.14569/IJACSA.2026.0171091 PDF

92

Federated Credit Card Fraud Detection Under Simulated Operating Conditions: Encrypted Aggregation, Partial Participation and Byzantine Resilience

Author 1: El Mehdi Adnani Author 2: Mohammed Erritali

Credit card fraud is rare, costly and continuously evolving, and the data required to detect it are distributed across institutions that cannot share them. This study presents a privacy-preserving federated framework in which several banks jointly train a fraud detector without exchanging raw transactions, and it quantifies the cost of… Read full abstract & cite →

Anomaly detection byzantine robustness credit card fraud detection federated learning homomorphic encryption

Abstract doi.org/10.14569/IJACSA.2026.0171092 PDF

93

Parallel BFS for Large-Scale Graph Processing and Smart Device Computing Frameworks

Author 1: Esra’a Alhenawi Author 2: Abdulla Aburomman Author 3: Omar Alidmat Author 4: Hebatullah Khattab

Large-scale graph processing and analysis have become crucial problems in the big data era, affecting a variety of fields like social networks, road, and web networks. Traditional Breadth-First Search (BFS) algorithms frequently scale inefficiently due to memory constraints and computational bottlenecks. This paper presents a parallel version of the BFS… Read full abstract & cite →

Parallel parallel breadth first search algorithm Hadoop concurrency smart device computing

Abstract doi.org/10.14569/IJACSA.2026.0171093 PDF

94

Read-First and Role-Gated Context Federation for Multi-Agent LLM Systems: A Controlled Multi-Model Comparison with Point-to-Point Messaging

Author 1: Ryan Valdivia-Segovia Author 2: Adrian Mamani-Quispe Author 3: Karim Guevara Author 4: Norka Bedregal-Alpaca

Multi-agent systems based on large language models (LLMs) commonly rely on point-to-point message-passing, which increases integration complexity and leaves fine-grained access control and traceability to additional mechanisms. This study evaluates the Graph Context Protocol (GCP), a read-first context-federation approach with role- and capability-based access control and native per-read provenance. GCP… Read full abstract & cite →

Large language models access control provenance agent containment software coupling

Abstract doi.org/10.14569/IJACSA.2026.0171094 PDF

95

AV-CACNet: A Cross-Attention Consistency Network for Audio-Visual Deepfake Detection in Cybersecurity

Author 1: Assem Turarbek Author 2: Sharipa Temirgaziyeva Author 3: Ahmet Sakir Dokuz Author 4: Alina Azanova Author 5: Nazgul Mametzhanova

The rapid development of audiovisual deepfake technologies has made the automatic verification of digital content authenticity a pressing issue, as methods relying solely on video or audio may fail to fully characterize complex forms of multimodal manipulation. This study proposes the AV-CACNet (Audio-Visual Cross-Attention Consistency Network) architecture, designed to jointly… Read full abstract & cite →

Audio-visual deepfake deepfake detection multi-modal learning cross-attention cybersecurity

Abstract doi.org/10.14569/IJACSA.2026.0171095 PDF

96

NA-CoLR: A Noise-Aware, Coherent and Self-Auditing Likelihood-Ratio Engine for Forensic Event Attribution Under Erroneous Reference Labels

Author 1: Gaseb Alotibi

Forensic examiners are increasingly expected to report the weight of digital evidence as a likelihood ratio (LR) rather than as a class decision. Three obstacles block this transition: calibration methods assume that reference labels are correct, although forensic reference labels are reconstructed and erroneous; multiclass problems are handled by independently… Read full abstract & cite →

Likelihood ratio calibration log-likelihood-ratio cost label noise empirical lower and upper bound evidence evaluation digital forensics uncertainty quantification

Abstract doi.org/10.14569/IJACSA.2026.0171096 PDF

97

When Offline Performance Misleads: A Leakage Audit of Completed-Session Features for Cart-Abandonment Prediction

Author 1: Babak Rad Jahanbani Author 2: Ahmet Sayar

Cart-abandonment models are often evaluated on session summaries created only after sessions end. Such tables can violate the information boundary of an intervention-time prediction task and inflate offline performance. This study audits 43,917 cart sessions reconstructed from the public Retail Rocket clickstream dataset. Feature-lineage and arithmetic-closure tests reveal two leakage… Read full abstract & cite →

Cart abandonment clickstream analytics data leakage temporal leakage offline evaluation point-in-time evaluation feature validity e-commerce

Abstract doi.org/10.14569/IJACSA.2026.0171097 PDF

98

Stock Price Forecasting with Technical Indicators: Comparing Windowed and Non-Windowed Deep Learning Models on the Saudi Stock Market

Author 1: Abdullah Almusned Author 2: Mohammad Mehedi Hassan Author 3: Muhammad Al-Qurishi Author 4: Bader Alkhamees

Stock price forecasting increasingly uses recurrent neural networks with sliding-window sequences. However, whether explicit temporal windowing improves next-day prediction when technical indicators already encode recent market behaviour remains unclear, particularly in the Saudi stock market. This study aimed to determine how temporal input design affects LSTM, hybrid LSTM-GRU, and MLP… Read full abstract & cite →

Stock price prediction deep learning long short-term memory sliding window technical indicators Saudi stock market

Abstract doi.org/10.14569/IJACSA.2026.0171098 PDF

99

X-ITFormer: Balanced Fusion and Multilevel Auditing for Multimodal Time-Series QA

Author 1: Leon Felipe Davis Coropuna Author 2: Percy Maldonado Quispe Author 3: Ana Maria Cuadros Valdivia

LLMs help narrow the semantic gap in time-series analysis by translating numerical patterns into natural-language descriptions. However, their adoption in critical domains is hindered by high computational requirements (models of 7B parameters or more) and by the difficulty of verifying whether predictions rely on the physical signal or on spurious… Read full abstract & cite →

Time series analysis multimodal learning multi-modal fusion question answering explainable AI small language models counterfactual explanations green AI

Abstract doi.org/10.14569/IJACSA.2026.0171099 PDF

100

A Novel Mobility-Aware Clustering Protocol for Internet of Vehicles Using Modified Traffic Jam Optimizer

Author 1: Wei Xia Author 2: Hanchuan Tong Author 3: Jihong Xie

The Internet of Vehicles (IoV) enables intelligent transportation through large-scale vehicular connectivity, but its high mobility and rapidly changing topology severely degrade communication stability and scalability. Clustering has emerged as an effective solution for reducing channel contention and improving network efficiency; however, existing methods often fail to maintain stable cluster… Read full abstract & cite →

Internet of Vehicles mobility-aware clustering traffic jam optimizer swarm intelligence

Abstract doi.org/10.14569/IJACSA.2026.01710100 PDF

101

AI-Driven Security Intelligence Framework for Knowledge Protection in Cloud Ecosystems

Author 1: Julius Olusegun Oyelami Author 2: Azleena Mohd Kassim Author 3: Nor Athiyah Abdullah Author 4: Mohd Heikal Husin Author 5: Nor Shamira Sabri

Cloud-based knowledge sharing creates a need to coordinate security controls with the protection of organizational knowledge. This study proposes a Security Intelligence Frame-work (SIF) linking computational intelligence (CI) and intelligent agents (IA) to artificial intelligence (AI), and AI to penetration testing (PT), cloud security (CS), and the secure systems development… Read full abstract & cite →

Artificial intelligence computational intelligence intelligent agents cloud security knowledge protection blockchain security

Abstract doi.org/10.14569/IJACSA.2026.01710101 PDF

102

Dynamic-Aware Point-Line Visual SLAM with Loop-Consistent Semantic Mapping

Author 1: Bo Jiang Author 2: Mengcheng Xu Author 3: Chenglong Li Author 4: Xi Wen Author 5: Xuanzhen Chen

Simultaneous localization and mapping (SLAM) plays an important role in autonomous vehicles and Unmanned Aerial Vehicles (UAVs) when satellite navigation is unreliable or unavailable. However, its performance often degrades in complex scenes with moving objects, weak texture, and map inconsistencies caused by loop closure. To address these issues, this paper… Read full abstract & cite →

Visual SLAM UAV Navigation point-line fusion dynamic environments low-texture scenes semantic mapping loop closure

Abstract doi.org/10.14569/IJACSA.2026.01710102 PDF

103

How Portable is ORM Query Performance? Evidence from Four Database Systems

Author 1: Mahmoud Nasr Author 2: Mohammed El-Ramly Author 3: Desoky Abdelqawy

A query that runs efficiently through an object-relational mapper (ORM) on one database may become a bottleneck on another. Such differences are easy to miss when ORM performance is summarized by a single overhead figure. This study examines how reliably performance carries across database systems by comparing Django and SQLAlchemy… Read full abstract & cite →

Object-relational mapping database benchmarking TPC-H TPC-C reproducibility measurement error

Abstract doi.org/10.14569/IJACSA.2026.01710103 PDF

104

Enhanced Adaptive Monte Carlo Localization with Iterative Closest Point Scan Matching for Agricultural Mobile Robots

Author 1: Brian Lai Lap Hong Author 2: Mohd Azri Mohd Izhar Author 3: Norulhusna Ahmad

Accurate localization of agricultural mobile robots is difficult in plantations because repeated rows of crops or trees provide limited distinctive geometry. Previously, an Adaptive Monte Carlo Localization (AMCL) refined by the Normal Distributions Transform (NDT) was proposed for two-dimensional Light Detection and Ranging (LiDAR) localization to improve the localization accuracy… Read full abstract & cite →

Adaptive monte carlo localization iterative closest point 2D LiDAR agricultural robotics scan matching pose estimation precision agriculture

Abstract doi.org/10.14569/IJACSA.2026.01710104 PDF

105

A Multi-Objective Optimization Algorithm for Batching Tasks in MEC

Author 1: Xuemei Yang Author 2: Junjun Zhang

For multi-source data fusion analysis applications in Multi-access Edge Computing (MEC) systems, an edge node processes the tasks offloaded by multiple devices within the same time window in the form of task batches. Batching tasks amortizes the computational overhead on the edge, but it also prolongs task waiting time and… Read full abstract & cite →

MEC batching tasks throughput maximization energy consumption minimization

Abstract doi.org/10.14569/IJACSA.2026.01710105 PDF

106

A Cross-Disease Comparative Analysis of CNN, Transformer, Hybrid, and Self-Supervised Models for Ocular Disease Classification

Author 1: Hazem Y. Al-Sebaay Author 2: Hadeer El-Saadawy Author 3: Maryam N. Al-Berry Author 4: Safaa El-Sayed Amin

Deep learning architecture comparisons in ophthalmic imaging are almost always conducted on a single disease and dataset, leaving open whether an architecture’s apparent advantage is a genuine, transferable property or an artifact of that one task. This study presents a controlled comparison of nine architectures spanning four families (CNN, Transformer… Read full abstract & cite →

Ophthalmic image classification vision transformers self-supervised learning glaucoma age-related macular degeneration cataract statistical validation architecture comparison

Abstract doi.org/10.14569/IJACSA.2026.01710106 PDF

107

Hierarchical Classification of Arabic Media: Employing a Two-Task BERT Model

Author 1: Aisha Blfgeh Author 2: Hanen Himdi Author 3: Rehab Qarout Author 4: Fatmah Assiri Author 5: Amjad Qashlan

Thousands of news articles are published on the web daily, causing the need to filter and extract knowledge on specific topics. With the increasing proliferation of digital content, automatically differentiating between informative news reporting and opinion-based columns has become essential to support fact-checking, content moderation, and journalistic integrity. Although previous… Read full abstract & cite →

Arabic text classification AraBERT BERT Hierarchical classification natural language processing news article classification opinion column detection transformer models

Abstract doi.org/10.14569/IJACSA.2026.01710107 PDF

108

Retrieval Familiarity for Selective Referral in Dermoscopic Classification: An External Evaluation Under Distribution Shift

Author 1: Pronab Kumar Mondal Author 2: Subrata Kumar Das Author 3: Sumon Kumar Debnath Author 4: Indrani Mandal Author 5: Md. Selim Al Mamun

This study audits retrieval familiarity as a post-hoc accept–refer signal for dermoscopic classification under dataset shift. Mean Top-5 familiarity was defined as the average cosine similarity between a query and its five nearest cases in a lesion-disjoint HAM10000 reference bank. Retrieval depth and the operating threshold were selected on HAM10000… Read full abstract & cite →

Dermoscopy selective classification selective referral distribution shift image retrieval

Abstract doi.org/10.14569/IJACSA.2026.01710108 PDF

109

Adaptive Hint-Level Selection in Intelligent Tutoring Systems: A Double DQN Study of Policy Behavior and Assistance Trade-Offs

Author 1: Jingwen Dong Author 2: Siti Nurulain Mohd Rum Author 3: Teh Noranis Mohd Aris Author 4: Khairul Azhar Kasmiran Author 5: Raihani Mohamed

Intelligent Tutoring Systems must determine not only whether assistance is needed, but also how much support should be provided during problem solving. This study investigates adaptive hint-level selection as a sequential decision problem in simulated mathematics tutoring. The action space consists of Strong Guidance, Partial Guidance, and Minimal Scaffold, and… Read full abstract & cite →

Adaptive hint selection instructional assistance reinforcement learning Double Deep Q-Network (DDQN) intelligent tutoring systems

Abstract doi.org/10.14569/IJACSA.2026.01710109 PDF

110

Path-Aware Heterogeneous Graph Reinforcement Learning for Green Light Optimal Speed Advisory

Author 1: Wei Zhao Author 2: Lingchao Wang Author 3: Qiang Ai

Green light optimal speed advisory (GLOSA) needs to jointly balance stopping, emissions, travel efficiency, and advice-update overhead. To address the difficulty of jointly optimizing the path traffic state and update frequency in multi-intersection GLOSA, this paper proposes a path-aware heterogeneous graph reinforcement learning method for green-wave speed advice. A heterogeneous… Read full abstract & cite →

Green Light Optimal Speed Advisory (GLOSA) reinforcement learning heterogeneous graph graph attention network speed advisory semi-Markov decision process

Abstract doi.org/10.14569/IJACSA.2026.01710110 PDF

111

Regime-Switching ESG Portfolio Optimization: Integrating Temporal Fusion Transformers and Reinforcement Learning for Enhanced Risk-Adjusted Returns

Author 1: Abdul Kadar Muhammad Masum Author 2: Md. Abul Kalam Azad Author 3: Chanda Rani Debi Author 4: ABM Yasir Arafat Author 5: Ramona Birau Author 6: Virgil Popescu Author 7: Gabriela Ana Maria Lupu (Filip) Author 8: Roxana Mihaela Nioata (Chireac) Author 9: Stefan Margaritescu

Integrating ESG criteria into institutional portfolios creates tension with financial objectives under changing market conditions. This paper proposes an architecture combining Temporal Fusion Transformer (TFT)–XGBoost forecasts, Gaussian Hidden Markov Model (HMM) inference, Proximal Policy Optimization (PPO), and constrained portfolio refinement. An executed annual chronological experiment separately evaluates a tractable allocation… Read full abstract & cite →

ESG investing reinforcement learning temporal fusion transformer Hidden Markov Model portfolio optimization regime detection multi-objective optimization ESG Beta

Abstract doi.org/10.14569/IJACSA.2026.01710111 PDF

112

Code Size or Code Shape? Re-examining the Complexity and Vulnerability Relationship in Human and AI-Generated Code

Author 1: Sultan S. Alqahtani

Software security research has traditionally associated greater code complexity with higher vulnerability risk. However, existing evidence is largely based on human-written code and often conflates structural complexity with code size. This study examines whether this assumption holds for AI-generated code by separating code size from size-normalised structural shape. This study… Read full abstract & cite →

Code complexity software vulnerabilities AI-generated code large language models static analysis CWE empirical software engineering

Abstract doi.org/10.14569/IJACSA.2026.01710112 PDF

113

Lightweight Federated Deep Learning for COPD Exacerbation Risk Prediction Using Multimodal Ubiquitous Health Sensing

Author 1: Riad Alharbey

Chronic obstructive pulmonary disease (COPD) exacerbations are difficult to anticipate because deterioration develops across heterogeneous respiratory, physiological, behavioral, environmental, and clinical signals. This study presents FedSqueeze-HGRUViTNet, a lightweight federated deep learning model for COPD exacerbation-risk prediction using multimodal wearable and harmonized clinical information. The model integrates depthwise separable convolution for… Read full abstract & cite →

COPD exacerbation prediction federated learning wearable health monitoring multimodal learning ubiquitous sensing deep learning

Abstract doi.org/10.14569/IJACSA.2026.01710113 PDF

114

Reliability Analysis and Scene-Aware Calibration of Federated Learning for Long-Tailed Autonomous Driving Scenes

Author 1: Song Wen Author 2: Shuling Yin Author 3: Jun Zhu Author 4: Xinying Wang

This study examines the predictive reliability of FedRare weather classification on BDD100K across 18 weather–time-of-day scenes. Labeled logits are calibrated centrally using 2,653 observations and evaluated on 6,188 observations from a frozen scene-stratified split. In the primary analysis, Predicted Scene-Aware Temperature Scaling (Predicted-SATS) reduces negative log-likelihood (NLL) from 0.7108 under… Read full abstract & cite →

Autonomous driving confidence calibration long-tailed distribution temperature scaling predictive reliability

Abstract doi.org/10.14569/IJACSA.2026.01710114 PDF

115

A Digital Decision Support Framework Integrating SARIMA and Intelligent Algorithms for Proactive Phytopathogen Forecasting: A Case Study on Late Blight

Author 1: Tamer Daghma Author 2: Maryam Hazman Author 3: Mostafa Ali

Climate change-induced variability in meteorological patterns significantly escalates systemic agricultural vulnerabilities by altering phytopathogen dynamics, rendering traditional reactive crop management strategies insufficient. To address these challenges, this paper presents the design, development, and validation of an integrated, climate-driven Intelligent Decision Support System (IDSS) that harmonizes statistical time-series forecasting with computerized… Read full abstract & cite →

Decision Support Systems (DSS) SARIMA climate change plant disease prediction knowledge-based systems Potato Late Blight smart agriculture Egypt

Abstract doi.org/10.14569/IJACSA.2026.01710115 PDF

116

A Neuro-Causal Fusion Architecture for Outlier-Robust ESG Forecasting and Policy Simulation

Author 1: Abdul Kadar Muhammad Masum Author 2: Md. Abul Kalam Azad Author 3: Najmus Saadat Author 4: Ramona Birău Author 5: Virgil Popescu Author 6: Ștefan Mărgăritescu Author 7: Roxana Mihaela Nioata (Chireac) Author 8: Cristina Sultanoiu (Patularu)

The global transition toward mandatory sustainability governance has exposed a critical "prediction-action" gap, where conventional black-box models fail to provide actionable strategic roadmaps. Current ESG forecasting is further compromised by the systemic risk of corporate greenwashing and the structural inability of standard machine learning to handle the heavy-tailed anomalies inherent… Read full abstract & cite →

ESG risk prediction neuro-causal fusion digital twin huber loss fsQCA sustainable finance

Abstract doi.org/10.14569/IJACSA.2026.01710116 PDF

117

Geometry-Aware Aggregation for Federated Learning: Down-Weighting Label-Corrupted Clients via Latent Class Prototypes

Author 1: Brahim CHEIKH SOUFI Author 2: Ridha Ejbali

FedAvg primarily weights client updates based on the volume of data for each participant, without considering the reliability of local labels. This approach can be problematic when a large client has unreliable labels, as its contribution remains significant only because of the number of samples it contains. To address this… Read full abstract & cite →

Federated learning label noise non-IID data class prototypes robust aggregation representation learning

Abstract doi.org/10.14569/IJACSA.2026.01710117 PDF

118

Investigating the Effect of the Data Partitioning Protocol on Automatic Alzheimer’s Disease Detection from Brain MRI Using Deep Learning

Author 1: Fatima Zahra Salmam Author 2: Sanae Belfrouh Author 3: Lamya Ouahmidi Author 4: Rahhal Errattahi

Computer vision and deep learning have been widely used to automatically detect and analyze many diseases from medical images. Some of these include tumor detection, diabetic retinopathy classification, nodules classification, and the screening of thoracic pathologies. In this work, we are investigating the application of deep learning techniques to the… Read full abstract & cite →

Alzheimer’s disease brain MRI ADNI deep learning transfer learning data leakage class imbalance

Abstract doi.org/10.14569/IJACSA.2026.01710118 PDF

119

XGBoost-Based Classification of Electromyography Signal for Muscle Contraction and Relaxation: A Correlation-Guided Feature Analysis Approach

Author 1: Nashreen Hilman Author 2: Nor Aziyatul Izni Author 3: Mohd Azizi Abdul Rahman Author 4: Mohamad Azlan Mohamed Shapie Author 5: Nur Ilyana Ismarau Tajuddin

Analyzing electromyography (EMG) signals has become one of the important methods in biomedical engineering for researching neuromuscular function that is non-invasive, with applications such as robotic rehabilitation, prosthetics, and human-machine interface design. Despite significant improvements in EMG pattern recognition, robust classification is still constrained by feature redundancy, the models’ weak… Read full abstract & cite →

Electromyography EMG classification XGBoost correlation analysis muscle contraction neuromuscular signal processing

Abstract doi.org/10.14569/IJACSA.2026.01710119 PDF

120

Multi-Source Rice Disease Detection Using Fine-Tuned Swin-Tiny with Grad-CAM Explainability

Author 1: Engy Youssef Author 2: Wael Hassan Gomaa Author 3: Hagar Mohamed

Rice (Oryza sativa) is a staple food for approximately half of the world’s population, which makes rapid and accurate detection of rice diseases essential. Most existing rice disease classifiers are trained and evaluated on images collected in a single region, so little evidence is available on their behavior across acquisition… Read full abstract & cite →

Rice disease classification deep learning transfer learning dataset fusion Swin Transformer EfficientNet Grad-CAM

Abstract doi.org/10.14569/IJACSA.2026.01710120 PDF

121

Variance-Based Stratified Undersampling with Capacity-Aware Sampling Control for Imbalanced Tabular Classification

Author 1: Mohd Hafiz Zakaria Author 2: Jafreezal Jaafar Author 3: Irving V. Paputungan Author 4: Said Jadid Abdulkadir

Minority class instances in imbalanced tabular data are routinely misclassified, and existing undersampling methods offer little principled basis for deciding which majority samples to discard, how many to remove, and when reduction should stop. We propose Variance-Based Undersampling (VBU), a tuning-free data-level method for binary imbalanced tabular classification that sets… Read full abstract & cite →

Class imbalance undersampling variance-based sampling stratified sampling data-level resampling tabular data classification minority class detection machine learning

Abstract doi.org/10.14569/IJACSA.2026.01710121 PDF

122

Federated Learning with TensorFlow Federated for Soil Classification in Smart Agriculture

Author 1: Marwen Ghabi Author 2: Sofiane Khalfallah Author 3: Hela Ltifi

In the field of smart agriculture, the use of agricultural data plays a key role in optimizing practices and improving decision-making. However, the centralization of this data raises significant challenges in terms of confidentiality and protection of information from farms. To address these challenges, Federated Learning (FL), implemented with TensorFlow… Read full abstract & cite →

Federated learning tensorflow federated artificial intelligence smart agriculture Machine Learning as a Service (MLaaS) microservices architecture soil classification data privacy predictive models scalability resilience

Abstract doi.org/10.14569/IJACSA.2026.01710122 PDF

123

A Controlled Benchmark of Active Learning Strategies for Class-Imbalanced Waste Detection on SortWaste with YOLO

Author 1: Thanh Thong Le Author 2: Tuyet Nhi Chu Author 3: Manh Hung Nguyen

Bounding-box annotation constrains waste detection under class imbalance and video redundancy. This benchmark compares Random, Plug and Play Active Learning (PPAL), Rarity-Aware, and an entropy/diversity strategy (Entropy) on SortWaste with YOLOv8n over six paired seeds and 5%–20%labeled-image budgets. The primary metric is normalized area under the budget curve (nAUBC) of… Read full abstract & cite →

Active learning object detection YOLO class imbalance waste detection image-budget efficiency

Abstract doi.org/10.14569/IJACSA.2026.01710123 PDF

124

AI-Assisted News Production in Media and Journalism: A Systematic Literature Review

Author 1: Ahmed Chebli Alami Author 2: Hajar Chnok Author 3: Aya Loudiyi Lamghafri Author 4: Abdeslam El Harraj Author 5: Soumia Ziti

Artificial intelligence (AI) is reshaping news production by enabling faster workflows, scalable content generation, and new forms of human–AI collaboration. However, the evidence supporting its effective and ethical use remains fragmented. This study presents a protocol-driven systematic literature review of AI-assisted news production published between 2015 and 2025, following the… Read full abstract & cite →

AI-assisted journalism automated news production ethical governance of AI human–AI collaboration newsroom integration systematic literature review

Abstract doi.org/10.14569/IJACSA.2026.01710124 PDF

125

A Blockchain-Governed Clinical Decision Support Framework with Federated Learning and Explainability

Author 1: Yukta Vishnoi

Clinical Decision Support Systems (CDSS) are widely used to support diagnosis, risk prediction, and treatment planning. However, many existing systems are limited by institution-centric data handling, weak auditability, and a lack of clear explanations at the point of clinical decision-making. Although blockchain-based federated learning approaches improve the security of model… Read full abstract & cite →

Clinical Decision Support System (CDSS) federated learning Explainable AI (XAI) Agentic AI

Abstract doi.org/10.14569/IJACSA.2026.01710125 PDF

126

Multi-Class Brain Tumor Classification from MRI Images Using Pre-Trained CNNs: Feature Representation and Uncertainty Analysis

Author 1: Mousa Alalhareth

Brain tumor classification based on Magnetic Resonance Imaging (MRI) is a challenging task due to the fact that there are many kinds of brain tumors that have quite similar visual features, and there are many different variations in individual cases of the same kind of brain tumor. The aim of… Read full abstract & cite →

Brain tumor deep learning models feature extraction magnetic resonance imaging

Abstract doi.org/10.14569/IJACSA.2026.01710126 PDF