NPS Australia Submission System
  • 0

    No products in the cart.

Paper Type: Full Research Paper

ViT-Sign: An Explainable Vision Transformer for Sign Language Recognition

Over 70 million deaf people worldwide frequently use sign language. However, the challenges of real-time implementation and interpretation have limited automated sign language recognition systems. CNN-based older techniques frequently overlook subtle features that are crucial for differentiating between similar motions. This research examines the recognition of 37 distinct sign language classes, which include A to Z, 0 to 9, and a space. We utilized a dataset of 55,500 photos for this. Our recommended method, ViT-Sign, uses a Tiny version of Vision Transformer. It has only 5.5 million parameters, and we evaluate its performance using accuracy, precision, recall, and F1-score. Our model achieves 99.42\% accuracy, 99.44\% precision, 99.42\% recall, and a 99.42\% F1-score. We also employ Grad-CAM visuals to make it clear that our model concentrates on the correct regions of the hands rather than random background information. Experiments show that Vision Transformers can efficiently record spatial relationships and acquire discriminative representations of gestures. The proposed method can be utilized to create successful communication systems that allow hearing-impaired people to communicate more easily.

Banana Leaf Disease Classification Using Explainable Deep Learning For Farmer-Oriented Advisory System

While banana is an important crop in the agricultural sector of South Asia, fungal and bacterial leaf diseases continue to cause substantial losses and pose a threat to the livelihoods of farmers in the region. The current method of field diagnosis is mostly manually, which is subjective and slow, inconvenient in the field. This paper proposes a rigorous and explainable deep learning framework for the automated banana leaf disease classification that overcomes three major drawbacks of previous works: 1) the lack of comparison of multiple models in similar experimental setups, 2) the lack of statistical validation, and 3) the absence of a connection between the model prediction and the farmer’s guidance. We test the three architectures: a lightweight Custom CNN, ResNet- 50 and EfficientNet-B0, on the publicly available BananaLSD dataset, based on a standardized preprocessing and augmentation pipeline. Stratified 5-fold cross validation is used for assessing the performance and paired t test is used to check the statistical significance of the difference observed. The best classification accuracy of EfficientNet-B0 is 99.69%, the mean cross-validation accuracy is 99.36%, and the standard deviation is the lowest (0.10), showing very high stability. For transparency, we integrate four explainability methods: Grad-CAM, Grad-CAM++, Score- CAM and Layer-CAM, which result in uniform lesion localized heatmaps that validate biologically meaningful feature focus. Finally, an intelligent advisory module translates the predictions in actionable management recommendations directly deployable by agricultural extension services. The proposed framework is built upon predictive excellence, interpretable decision making, and real-world usability in one seamless pipeline, which is critical for advancing precision agriculture.

Digital Financial Inclusion for Resilient Digital Ecosystems in Fiji and the Pacific Islands

The research contributes by systematically identifying and synthesising the dual role of emerging technologies, specifically human -centric artificial intelligence (AI) and decentralised digital digital systems, in shaping financial inclusion outcomes. It demonstrates how these technologies simultaneously act as enablers of access and efficiency, while also introducing new barriers related to digital literacy, transparency and trust.

An Explainable Multi-Objective Genetic Algorithm Framework for Healthcare Fraud Detection

• A MOGA-based framework is proposed for healthcare
fraud detection, which simultaneously optimizes classifi
cation performance and feature subset size.
• The framework integrates LIME to enable instance-level
interpretability, facilitating a deeper understanding of
model predictions.
• A comprehensive comparative analysis is conducted to
evaluate model performance and interpretability before
and after feature selection.

Multi-Criteria Decision-Support System for Sustainable Management of the Aral Sea Basin under Uncertainty

— The Aral Sea Desiccated Basin faces one of Central Asia’s most severe ecological and socio-economic crises, driven by decades of unsustainable water use and climate change. Developing sustainable management strategies for this region remains complex due to uncertain, incomplete, and often conflicting environmental data. This study presents a hybrid Multi-Criteria Decision-Support System (MCDSS) that integrates Intuitionistic Fuzzy Sets (IFS) with Multi-Criteria Decision-Making (MCDM) techniques to address uncertainty and hesitation in environmental decisions. By incorporating membership, non-membership, and hesitation degrees, the proposed model enables a more realistic evaluation of management alternatives compared to traditional deterministic or fuzzy methods. To further enhance analytical capacity, a Large Multimodal Model (LMM) is introduced to process and fuse satellite imagery, numerical indicators, and expert textual inputs. The LMM supports cross-modal reasoning and improves the interpretation of soil salinity, vegetation degradation, and water resource distribution. A case study on water allocation and land rehabilitation in the Aral Sea Basin demonstrates that the LMM-enhanced IFS–MCDM framework improves decision robustness, adaptability, and transparency. The results suggest that the proposed system can effectively support data-driven, sustainable management strategies for ecologically vulnerable regions under uncertainty.

AI Driven Personalised Learning for Sustainable Higher Education in the Pacific

Personalised learning (PL) is increasingly promoted
as a strategy to enhance engagement and achievement in higher
education. Advances in adaptive systems, artificial intelligence, and
learning analytics enable tailored pathways and real-time feedback.
Meta-analyses report medium-to-large effects for intelligent
tutoring systems (ITS, g ≈ 0.6–0.7) and moderate gains for selfregulated
learning (SRL) interventions (d ≈ 0.69), while adaptive
platforms typically yield small-to-moderate improvements. Largescale
implementations, however, often produce only modest results
with considerable variance across contexts. This paper presents
a PRISMA-guided scoping review (2012–2025, n = 92 studies)
synthesising global evidence on PL modalities and applying
insights to Pacific higher education, where challenges of access,
equity, and cultural fit are acute. The contribution is twofold: first,
it explicitly links global effect-size evidence to equity outcomes
in low-resource contexts; second, it proposes an offline-first,
edge-based AI architecture tailored to Pacific higher education
to mitigate connectivity, privacy, and sustainability barriers.
The paper concludes with a context-sensitive research agenda
emphasising hybrid models, equity-first design, and governance
frameworks for sustainable implementation.

Decision-Making Framework for Sustainable Management of the Aral Sea Dried-Bottom under Uncertainty

The Aral Sea region has been facing severe ecological, social, and economic challenges due to large-scale water mismanagement and climate change impacts. Effective environmental decision-making in this area requires the integration of multiple uncertain and conflicting factors. Traditional decision-making approaches often fail to adequately handle the complexity and vagueness of such problems. This paper proposes a novel decision-making framework based on the theory of Intuitionistic Fuzzy Sets to address uncertainty in the management and ecological rehabilitation of the Aral Sea. The developed approach enables decision-makers to model hesitation and incomplete knowledge more effectively compared to classical fuzzy methods. A case study focused on sustainable water resource allocation and ecological restoration strategies in the Aral Sea basin demonstrates the applicability of the proposed model. The results show that the intuitionistic fuzzy approach provides greater flexibility and reliability in evaluating alternatives, thereby supporting more robust and sustainable environmental management decisions under uncertainty.

SSViT-4.0: A Self-Supervised Hybrid CNN-Transformer Framework for Industrial Visual Anomaly Detection

Detecting industrial visual anomalies remains a critical challenge in smart manufacturing due to scarce labeled defect samples and highly variable texture patterns. Existing anomaly detection (AD) and active learning (AL) approaches often struggle under adversarial conditions, as training typically relies on only normal, unlabeled data. This research proposes SSViT‑4.0, a self-supervised hybrid framework combining Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) for reliable, label-efficient anomaly detection in industrial images. A cross-hierarchical fusion module integrates global ViT self-attention with local CNN feature extraction, overcoming limitations of conventional independent feature streams or supervised fine-tuning. Self-supervised pretraining via reconstruction enables efficient representation learning without manual annotations, while patch-based embeddings and a lightweight anomaly scoring head allow accurate anomaly localization at inference. Experimental results on benchmark datasets (e.g., MVTec AD) demonstrate that SSViT‑4.0 surpasses CNN-only and Transformer-only baselines in detection accuracy and localization, maintaining real-time inference. The framework offers a scalable and efficient solution for automated visual inspection in Industry 4.0 environments.

Potential Use of Natural Fibre Reinforced Composites as Alternative Materials for Wind Turbine Blades – Short Review

With the rise and implementation of decarbonization policies at global level, there has been a rapid and increasing shift towards renewable energy sources. Wind energy has been one such alternative and greener source of energy which uses the wind turbine blades (WTBs) to harness and convert the kinetic energy of wind into electrical energy. On the downside, WTBs have poor recyclability attributes, which tend to be costly and complex, as they are manufactured from high-strength composite materials to meet their lightweight and intricate airfoil shape requirements. With many of the wind turbines approaching the end of their lifetime in service in the coming years, a surge in the amount of WTBs destined for the landfill is forecasted to surge in the upcoming decades. This could result in an ecological issue which could be worsened if sustainable measures are not considered and implemented in the medium and long term. This study aims to review the use of alternative sustainable materials, namely Natural Fibre Reinforced Composites (NFRCs), which could replace presently-used composites such as Glass Fibre Reinforced Composites (GFRCs) and Carbon Fibre Reinforced Composites (CFRCs). In the first part of this study, some typically used NFRCs and their key attributes for WTBs fabrication were reviewed. The potential of using fully biodegradable NFRCs for WTBs and their associated benefits were also discussed. Finally, the notable advantages and limitations observed when integrating NFRCs into modern WTBs were further discussed in this short review study. The findings of this study clearly show that the use of NFRCs in WTBs can largely enhance sustainability in the wind renewable energy sector while concurrently addressing some of the ecological factors associated with their disposal stage.

Trends in Cybersecurity Threats and Mitigation Strategies: A Decade of Global Evidence

This paper contributes to cybersecurity research by providing a decade-long analysis (2015–2024) and forecast analysis (2025-2030) of global cybersecurity threats, integrating statistical testing with trend visualization to evaluate both frequency and severity of incidents. Unlike prevailing assumptions of linear growth, the study demonstrates that while overall volumes have not increased significantly, specific attack types such as ransomware and phishing remain dominant and financially damaging. Furthermore, the comparative analysis of defense mechanisms reveals that incident response workflows, rather than specific technologies, largely determine resolution effectiveness. These insights offer evidence-based guidance for policymakers, practitioners, and organizations seeking to strengthen resilience through adaptive, data-driven cybersecurity strategies.

Web-based Explainable Machine Learning Model for Early-Stage Heart Disease Detection

• Proposed a web-based system using a Voting Ensemble (VE), a soft-voting classifier that combines K-Nearest Neighbors (KNN), Gradient Boosting (GB), and CatBoost with optimized hyperparameter for improved heart disease prediction.

• Evaluated model balance using 10-fold cross-validation with a held-out test set, ensuring robust performance and generalization.

• Integrated SHAP to give feature-level interpretability, enhancing clinical trust in model predictions.

• Developed a web-based application for real-time heart disease risk prediction, demonstrating potential for seamless clinical deployment.

XAI-Enhanced Hybrid Models for Effective Chronic Kidney Disease Prediction

Our key contribution lies in developing a hybrid predictive framework that combines machine learning, deep learning, and ensemble methods for highly accurate and interpretable Chronic Kidney Disease (CKD) detection. We addressed data imbalance through SMOTE, SMOTE-Tomek, and SMOTE-ENN, followed by SHAP-based feature selection via XGBoost, where top-ranked features were aggregated across all resampling strategies. Complementing this, statistical analysis with SPSS reinforced the identification of clinically significant features. Multiple machine learning classifiers, ensemble approaches, and advanced deep learning models—including Attention Autoencoder with XGBoost, TabNet, TabPFN, LightCNN, MLP, and DeepCrossNet—were systematically evaluated. DeepCrossNet achieved 97.38% accuracy, while the stacking ensemble attained 97.50% and Random Forest reached 97.71%. Furthermore, SHAP and LIME explanations emphasized GFR and serum creatinine as critical predictors, enhancing clinical trust. Visualization with t-SNE and UMAP confirmed class separability and detected ambiguous cases. Together, these contributions highlight the novelty of integrating SHAP-aggregated features, advanced resampling, and hybrid ML-DL ensembles to advance both accuracy and interpretability in CKD prediction.

Sustainable Development in Transition: Empirical Insights on Green Economy and Carbon Emissions from India

This study contributes by providing empirical evidence on the link between the green economy and carbon emissions in India, a country undergoing rapid economic transition. It fills a critical gap by integrating sustainability and emission dynamics in a single framework, offering insights that extend existing literature. The findings have strong policy relevance, supporting strategies for low-carbon growth and sustainable development pathways in emerging economies.

Women-led Social Entrepreneurship in Fostering SDGs: Myth or Reality? A Phenomenological Perspective

Abstract—The study explores women’s social entrepreneurial approaches that align with the United Nations’ Sustainable Development Goals, reflecting overall sustainable development in the context of Bangladesh, targeting achievement by 2030. The study adopted a phenomenological approach to reveal the lived experiences of six women social entrepreneurs’ journey towards social entrepreneurship. Protocol analysis, conducted through in-depth interviews, was analyzed through a phenomenological lens. Following Husserl’s epoche (bracketing) and eidetic reduction, it captures the essence (noema) and the act of experiencing (noesis) of underlying entrepreneurial approaches towards the SDGs. The findings of the study demonstrate nine key approaches and their alignment with the SDGs and sustainable development of Bangladesh. By examining lived experiences and interpreting realities, it seeks to provide policymakers with actionable insights to facilitate supportive arrangements for schemes, mentorship, and networking. Additionally, it aims to empower educators to serve findings as learning modules and to engage both national and international development organizations in actively involving social entrepreneurs as key partners in sustainability efforts. Future research endeavors will shed light on men and other supportive organizations, addressing the SDGs within the context of different countries.

Preventing Local Minima in Lyapunov-Based Control Scheme through Iterative Repulsive Potential Augmentation

This study introduces the Iterative Repulsive Potential Augmentation (IRPA), a motionplanning framework designed to overcome the local minima problem in Artificial Potential Fields. IRPA builds upon the Lyapunov-based Control Scheme (LbCS) by iteratively increasing the repulsive potential in regions where the robot becomes trapped.
This augmentation is repeated across successive iterations until the robot can successfully escape the local minimum and reach its goal. The approach provides a computationally efficient solution to the local minima problem while preserving the inherent efficiency and scalability advantages of the LbCS.

Comparative Analysis of Supervised Machine Learning Anomaly Classification on Active Power in Smart Grids.

The primary contributions of this study are as follows. First, it presents a focused investigation on active power–based anomaly classification in smart grid physical infrastructure by utilizing the physical attack subset of a cyber–physical dataset. Second, it conducts a systematic comparative evaluation of multiple supervised classifiers under physical attack scenarios, employing rigorous multi-class performance metrics such as precision, recall, F1-score, and confusion matrices. Third, it applies Random Forest–based feature-importance methods, including Gini impurity and permutation importance, to identify influential features, reduce dimensionality, and enhance interpretability. Finally, the study advances toward an explainability-driven and practically deployable anomaly classification and detection framework for smart grid monitoring, addressing limitations of prior works constrained by dataset scope, conceptual focus, or narrow evaluation metrics.

Signet: A Low-Cost Prosthetic Glove For Deaf And Mute Patients

Speech is the key to human existence. These are
not easily communicable in the speech-and-hearing-impaired
people. If you consider traditional alternatives like sign
language or even writing in a notepad, the other person also
has to be familiar with those forms of communication rather
than there being a real-time disconnect. In this paper, a low
cost touch-sensor-based assistive communication glove is
proposed, which differentiates between seven specific touches
across fingers and converts them into text messages displayed
on the LCD screen. The system incorporates capacitive touch
sensors, Arduino Nano microcontroller for computation, and
an LCD with an I2C module for display. The proposed design
is intended to provide a portable, low-cost, and user-friendly
solution improving the independence of the deaf and mute
society. Experimental results indicate that the proposed
system is efficient, easy, and ideal for routine daily
communication.

An Efficient Attention-based Deep Learning Model for Masked and Unmasked Face Recognition

• Proposed the Custom Aug-CNN-5, a CNN-based architecture optimized for recognizing masked and partially occluded faces.
• Integrated attention mechanisms (CBAM) and cosine annealing to improve learning efficiency and enhance discriminative power.
• Conducted a comprehensive evaluation on the larger customized VGGFace2 dataset, demonstrating high accuracy, generalizability, and fairness across varied conditions.

Modeling Stop–and–Go Wave Dissipation under Partial CAV Penetration using a Multi-Class CTM

This paper investigates the dissipation of stop-and
go traffic waves under partial penetration of connected and
autonomous vehicles (CAVs) using a multi-class cell transmission
model. A controlled numerical study was conducted along a
corridor with a localized bottleneck, with experiments varying
CAV penetration levels and smoothing intensities. The validated
results show that even moderate penetration of CAVs substan
tially improves traffic performance: unstable density patterns are
suppressed, shockwave propagation is attenuated, and total delay
decreases monotonically with increasing adoption. Fundamental
diagrams with triangular envelopes confirmed capacity gains
consistent with reduced effective headways, while regression of
the jam front demonstrated qualitative agreement between mea
sured and theoretical shockwave speeds. These findings indicate
that CAVs can deliver network-level benefits well before full
market penetration, supporting their role as a viable strategy
for congestion mitigation. The study contributes both a method
ological framework for evaluating CAV impacts in macroscopic
models and empirical insights to guide deployment and traffic
management policy.

Leveraging Large Language Models to Investigate the Decoy Effect in Route Choice Behavior

Route choice behavior is a cornerstone of transport
research, traditionally modeled under the assumption that trav
elers act as rational agents who maximize utility by trading off
time, cost, and reliability. However, behavioral economics shows
that real-world decision making often departs from rationality
due to systematic biases. One such bias is the decoy effect, which
occurs when the introduction of a dominated alternative increases
the attractiveness of another option. While widely studied in
consumer contexts, its role in transport decision making remains
underexplored. This paper presents a novel methodology that
leverages large language models (LLMs) to investigate the decoy
effect in route choice. Using ChatGPT-4o mini, we generated
textual framings of a dominated route alternative and employed
the model as a synthetic respondent to simulate route selections.
Four experimental conditions were tested: baseline (two routes),
decoy with neutral framing, decoy with positive framing, and
decoy with negative framing. Results demonstrate that the decoy
increased the relative attractiveness of the premium route, partic
ularly under positive framing, while negative framing attenuated
the effect. Synthetic responses closely aligned with predictions
from a multinomial logit model, confirming consistency with
behavioral theory. The findings illustrate that LLMs can serve
as both framing generators and behavioral simulators, offering
a rapid testbed for prototyping hypotheses in transport research.
This work establishes proof-of-concept evidence that LLMs can
complement traditional behavioral models, opening pathways
for future integration of artificial intelligence into the study of
systematic biases in mobility decisions.

Explainable Flight State Recognition: Toward Intelligent Pilot Training Systems

This study presents a novel framework that integrates explainable artificial intelligence (XAI) to enhance classification of flight states of an aircraft in the context of pilot training. By combining probabilistic outputs from Artificial Intelligence (AI) models, the approach identifies high-confidence and low-confidence predictions. Further, Shapley Additive Explanations (SHAP) technique is employed to uncover feature-level insights.
The main contribution of the study is to know how model identifies the ambiguous flight states and to know the most influential features in classifying these states. Moreover, to demonstrate how interpretable AI can support the development of intelligent, data-driven pilot training systems that are grounded in real flight data and model transparency.

Effectiveness of Forensic Tools in Extracting Web Browser Artifacts

Comprehensive Multi-Browser Analysis
Scenario-Based Experimental Design
Multi-Tool Evaluation
Evidence of Residual Artifacts in Privacy Modes

Predicting Online Delivery Adoption During COVID-19: A Machine Learning Approach

The significant research contribution of this study lies in its comprehensive analysis of the sociodemographic and behavioral factors influencing online delivery adoption during the COVID-19 pandemic, utilizing advanced machine learning techniques. By employing a structured preprocessing pipeline and comparing multiple models, the study identified LightGBM as the most effective classifier, achieving an accuracy of 97% and an F1-score of 0.96 for both classes, which underscores its capability to capture complex patterns in consumer behavior. The findings revealed that younger age, higher household income, and education level were the most influential predictors of increased delivery use, providing valuable insights for transportation planners and policymakers. This research not only enhances understanding of consumer behavior during a critical period but also offers a framework for future studies on urban logistics and service adaptation in response to evolving consumer needs.

LiteFakeNet: Efficient Deepfake Image Detection with Depthwise Separable Convolutions

The significant research contribution of this study is the development of LiteFakeNet, a novel lightweight convolutional neural network (CNN) designed for efficient deepfake image detection, achieving an accuracy of 95%, precision of 96%, and recall of 94% on the CIFAKE dataset, which consists of 120,000 images. LiteFakeNet utilizes depthwise separable convolutions to balance high performance with low computational and energy costs, featuring less than 83,000 parameters and only 0.16 million FLOPs, making it more efficient than existing models like MobileNet and ResNet50. This model not only addresses the urgent need for effective deepfake detection but also aligns with the principles of Industry 5.0 by promoting human-AI collaboration and sustainable technology.

StegoVision: Enhanced Video Steganography via Prewitt Edge Mapping and 3-XOR Secured LSB

The significant research contribution of this study is the development of an automated two-tier data concealment technique for video steganography that integrates Advanced Encryption Standard (AES) encryption with a robust steganographic method, enhancing data security and imperceptibility. This approach utilizes a 128-bit AES key for encrypting sensitive information, which is then embedded into video frames using a combination of the Least Significant Bit (LSB) method and a Prewitt pixel selection technique, achieving superior data invisibility while maintaining frame quality. The proposed model also incorporates a Fisher-Yates randomization method for frame selection, which increases the resilience and payload capacity of the steganographic process, and demonstrates improved performance metrics such as Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) compared to existing techniques.

Exploring the Research Landscape of DCT-Based Video Steganography: A Systematic Review

The systematic literature review highlights significant advancements in DCT-based video steganography, particularly through hybrid techniques that combine DCT with other methods like DWT and error correction codes (ECC) to enhance imperceptibility and robustness. Notable contributions include the use of deep learning models, such as CNNs, to maintain visual quality while embedding data, and the development of coverless techniques that do not alter the original video content. The review also emphasizes the importance of performance metrics like PSNR and MSE for evaluating the effectiveness of these techniques.

Hyperparameter-Tuned Logistic Regression for Early-Stage Breast Cancer Prediction using Explainable AI

• Design of HTLR model using optimized logistic regression applied to early stage breast cancer prediction.

• Integration of SHAP for model interpretability, supporting clinicians in extracting key features affecting malignant and benign predictions.

• Comparative analysis with baseline state-of-the-art models and existing systems, demonstrating that the proposed HTLR achieved 99.12% accuracy while providing transparent and clinically meaningful explanations.

Deep Learning-Based Forecasting and Analysis of Urban Air Quality Using LSTM and Statistical Methods
Structured Prompting and Multi-Agent Reasoning for Open-Source Financial Sentiment Analysis

• A novel structured prompting methodology that provides context-aware, declarative instructions to coordinate multi-agent collaboration, and an output aggregation mechanism that assigns greater weight to more confident
agent responses, improving reliability and interpretability.
• A manually annotated, domain-specific dataset designed to support uncontaminated, realistic performance evaluation.
• A resource-efficient system combining LLaMA 3.1 8B for specialized subtasks and LLaMA 3 70B for synthesis, balancing scalability and performance.

Analytical Framework for a Green Room: An Integrated Passive Cooling Approach for Sustainable and Climate Resilient Buildings

This research introduces the Green Room framework, the first closed-form model integrating four passive cooling strategies. By quantifying synergistic effects and embedding environmental drivers, it delivers a retrofit-ready, simulation-free solution capable of 41–44 °C cooling, establishing a scalable pathway toward sustainable, climate-resilient, and energy-efficient building design.

Accelerated Skin Prick Test (aSPT): A Low-Cost, AI-Guided Allergy Diagnostic System Using Microneedles and Image-Based Severity Scoring

Introduces a low-cost allergy diagnostic integrating a dual-layer microneedle patch with offline AI image analysis, enabling painless, rapid, and accessible testing under $4 per test with inference in <1s per allergen site

AI Regulation in the U.S.: Lessons from Commodity Futures Legislative Journey and a Perspective CFTC-NFA Model for AI

The contributions of this article are threefold: first, it presented commodity futures, an equally disruptive and controversial invention as AI, that has culminated chaotic but successful legislative journey. Futures’ legislative path can serve as a guide for regulating AI. Second, it extracted key success factors and lessons learned from futures regulation, including principle-based laws, a designated coordinating agency, and a supporting self-regulatory organization (SRO) that represents the industry. Third, it recommended a potential CFTC-NFA-type legislative framework for AI regulation.

Amanot-Net: A Resilient Framework for Proactive Threat Neutralization in Unstructured Public Spaces

This research presents Amanot-Net, a proactive security framework featuring a novel dual-channel communication system (SIP and LoRaWAN) to guarantee alert delivery in unreliable networks. Key contributions include the public release of ‘BD-Threats-v1’, a new dataset with over 73,000 images contextualized for Bangladesh , and strong empirical results: 0.89 mAP@0.5 , end-to-end latency under 450ms , and 97% alert delivery resilience via the LoRaWAN fallback channel.

Improving Interpretability in Lung Cancer Prediction through Explainable Ensemble Voting Approach

• Presented an Explainable Ensemble Voting Approach (EEVA) which combines Random Forest (RF), Decision Tree (DT) and Multi-Layer Perceptron (MLP) using soft voting, to forecast lung cancer.

• Utilized LIME and SHAP to generate local and global explanations, enhancing model transparency and interpretability.

• Achieved an accuracy of 92.86%, outperforming baseline methods and demonstrating strong potential for clinical application.

Security Dilemma in Software Development: a Case Study on Understanding Developer Priorities and Practices in Stack Overflow

Applications have advanced rapidly in recent years, taking software development to a whole new level. These advancements have led to a growth in the complexity of applications, cloud computing, and the Internet of Things (IoT). As a result of this improvement, software security has become a paramount concern for developers, but it has traditionally been overlooked. But now, security issues are evolving day by day. In this paper, we explore developer priorities and practices in term of security in Stack Overflow website, whether they implement their software with security in mind or not. We utilize the Stack Exchange Data Dump to collect and compile a dataset of questions and their corresponding answers related to security vulnerabilities, specifically those that include user-submitted programming code snippets, our analysis concentrates on security-related topics. The experimental results indicate that Python emerged as the most commonly used language for security code snippets across these topics. From the expanded sample of code snippets, several vulnerabilities are flagged for security issues.

Deep Learning Based Bioimpedance Spectroscopy in Organic Communications: A Feasibility Analysis

The research contributions of this paper are:
-An SFD-BIS framework for vegetation-based OANs.
-A link established between channel response and sugarcane sucrose content using finite element analysis (FEA).
-A deep learning-based sucrose content estimator developed as a use case for the SFD-BIS framework.

AI -Based Tutor for Visually Impaired Students

Visually impaired students typically face intense
difficulties when handling electronic learning materials such as PDF
textbooks, academic papers, and study guides. Traditional assistive
technologies like screen readers have the basic text-to-speech
functionality without enabling smart interaction and understanding
document structure. This paper outlines the design and
implementation of an AI-Based Tutor system to enhance the
learning experience for visually impaired students. The system
integrates Optical Character Recognition (OCR) to extract
structured text from diverse PDF structures, including accurate line
and section detection. The system has high-quality Text-to-Speech
(TTS) capabilities for natural-sounding speech, and a Speech-toText (STT) functionality for real-time voice queries and
instructions. Essentially, the system employs leading-edge Natural
Language Processing (NLP) models such as GPT or Gemini to
provide intelligent question answering, context-based
conversations, and content summarization. A voice-enabled
interface affords hands-free and intuitive control of the learning
process. Preliminary evaluation with blind students shows increased
accessibility, comprehension, and interaction efficacy compared to
existing solutions. The current paper contributes a new, integrated
solution that provides blind students greater independence and
control over their studies

AI -Based Tutor for Visually Impaired Students

Visually impaired students typically face intense
difficulties when handling electronic learning materials such as PDF
textbooks, academic papers, and study guides. Traditional assistive
technologies like screen readers have the basic text-to-speech
functionality without enabling smart interaction and understanding
document structure. This paper outlines the design and
implementation of an AI-Based Tutor system to enhance the
learning experience for visually impaired students. The system
integrates Optical Character Recognition (OCR) to extract
structured text from diverse PDF structures, including accurate line
and section detection. The system has high-quality Text-to-Speech
(TTS) capabilities for natural-sounding speech, and a Speech-toText (STT) functionality for real-time voice queries and
instructions. Essentially, the system employs leading-edge Natural
Language Processing (NLP) models such as GPT or Gemini to
provide intelligent question answering, context-based
conversations, and content summarization. A voice-enabled
interface affords hands-free and intuitive control of the learning
process. Preliminary evaluation with blind students shows increased
accessibility, comprehension, and interaction efficacy compared to
existing solutions. The current paper contributes a new, integrated
solution that provides blind students greater independence and
control over their studies

Energy-Efficient IoT and Cloud Framework for Accurate and Scalable Hotel Occupancy Monitoring

The main contributions of this paper are summarized as follows:
• Development of a low-cost IoT-based hotel room occupancy detection system using radar sensing for both static and dynamic presence.
• Cloud-based storage and visualization through Firebase and a Django dashboard, enabling real-time and historical monitoring.
• Energy-efficient and sustainable operation using deep sleep, Wi-Fi Manager, and OTA-enabled remote maintenance.
• Demonstration of scalability across domains beyond hospitality, including smart homes, healthcare, and intelligent building systems.

An Explainable AI-Based Ensemble Machine Learning Framework for Early-Stage Diabetes Prediction

• Introduced the Explainable Ensemble Learning Framework (EELF) a Voting Classifier, that integrates Logistic Regression, Random Forest, and K-Nearest Neighbors with optimized hyperparameters to predict diabetes.

• Incorporated SHAP and LIME to enhance model interpretability by identifying key feature contributions, thereby improving clinician trust in the decision-making process.

• Conducted a comparative analysis, where the EELF achieved an accuracy of 81.16%, demonstrating strong potential for clinical application.

An Approach to Validate References in Scholarly Articles using RoBERTa

The significant research contribution of this paper lies in proposing a semi-automatic digital system for validating references in scholarly articles using RoBERTa-based semantic similarity analysis.

Key contributions include:

Introducing a novel framework that leverages RoBERTa embeddings with K-similar search to verify references against cited works.

Overcoming BERT’s input length limitations by applying document segmentation and preprocessing strategies for handling long research papers.

Achieving higher accuracy (F1-score: 0.777) compared to BERT and SBERT, demonstrating the effectiveness of RoBERTa for contextual similarity in reference validation.

Reducing reliance on manual cross-checking and peer reviewers, thereby streamlining the academic publication process while preserving reference authenticity.

Image captioning: the application of deep learning to improve the feeling expression and translation process

An ‘emojian’ algorithm was developed to manage this procedure. The enhanced caption produced, is then translated through Google/Argosopentech translate API, to the particular natural language picked manually, or through the automatic reading of the apparatus. Eventually, the translated enhanced mixed caption is combined with the scanned image, derived through the initial stage of this process. To assess and verify the effectiveness of the Emojian deep learning algorithm, we scrutinized the complete word count in the given text caption parameter relayed to our algorithm, and the complete emoji count delivered by the algorithm, each time a positive result is posted.

IntelliGrid: A Hybrid CNN–LSTM Framework for Intelligent Fault Detection in Smart Grids Using PMU Data

The increasing penetration of renewable energy
and inverter-based distributed generation has introduced
significant challenges for fault detection in modern smart grids.
Traditional protection schemes often struggle to identify weak
or evolving fault signatures, leading to delayed fault isolation
and compromising system reliability. To address this gap, this
study introduces a hybrid deep learning framework that
integrates convolutional neural networks (CNNs) with long
short-term memory (LSTM) networks for intelligent fault
detection. The approach leverages phasor measurement unit
(PMU) image data, where CNN layers extract spatial fault
characteristics and LSTM layers capture temporal dynamics,
enabling a more comprehensive representation of fault
progression. Data preprocessing included normalization, class
rebalancing, and synthetic noise augmentation to ensure
robustness. Model performance was validated using stratified 5-
fold cross-validation, achieving 99% classification accuracy
while maintaining lower computational requirements compared
to CNN-only and ensemble-based baselines. Comprehensive
evaluation with ROC–AUC, PR–AUC, per-class accuracy, and
confusion matrices further demonstrated reliability and
interpretability. The findings highlight the potential of the
proposed method to enhance fault detection mechanisms,
contributing to improved grid stability, faster protection
response, and greater resilience of future smart energy systems

Real-Time Intrusion Detection in Smart EV Charging Networks Using Embedded Deep Learning

The increasing connectivity of Electric Vehicle Supply Equipment (EVSE) within smart grid networks has compounded the threat of cyber exposures, particularly Distributed Denial-of-Service (DoS) attacks. This paper introduces a lightweight, real-time intrusion detection system based on a hybrid deep learning structure that intermixes Transformer encoders and a Multilayer Perceptron (MLP) classifier. The proposed model uses kernel-level event logs to capture both temporal dependencies and high-dimensional feature interactions. A well-structured preprocessing pipeline includes feature leakage prevention, normalization, class balancing, and stratified cross-validation. This approach ensures data integrity and effective learning. Experimental evaluation on a real-world dataset confirms the model’s superior performance with 100% accuracy, precision, recall, and F1-score, and ideal ROC-AUC and PR-AUC scores. Furthermore, the framework is streamlined for deployment on resource-constrained edge devices to enable decentralized, on-device threat detection. These results accentuate the effectiveness and viability of the suggested solution in enhancing the cybersecurity posture of modern EV charging ecosystems

Data-Driven Condition Monitoring and Fault Detection of Power Transformers Using ML

Reliable transformer operation is critical for
minimizing downtime and ensuring power system stability.
Dissolved Gas Analysis (DGA) is the most widely used diagnostic
tool, yet ratio-based methods such as the Duval Triangle and
Key Gas Method often fail when signatures overlap or appear
at early fault stages. While machine learning has improved
accuracy, models remain vulnerable to noise, imbalance, and
overfitting. This paper proposes a CatBoost-based framework
that combines statistical and energy features of H₂, CO, C₂H₂,
and C₂H₄ gases with engineered ratios to capture complex intergas dependencies. With tuned hyperparameters, the model
achieved 97.6% overall accuracy and strong class-wise
performance: 98.9% (Normal), 94.7% (Partial Discharge),
93.9% (Low-Energy Discharge), and 89.0% (Low-Temperature
Overheating). Feature importance analysis identified H₂ and gas
ratios as key contributors, while training dynamics showed
rapid and stable convergence. The results demonstrate
robustness, interpretability, and efficiency, highlighting the
framework’s potential for real-time transformer fault detection
and improved power system resilience

Explainable Ensemble Machine Learning Framework for Accurate Detection of Thyroid Disorders

• We have introduced a highly developed Ensemble-Based Thyroid Disorder Detector (ETDD) that uses Random Forest, CatBoost, LightGBM, and XGBoost in sequence, and feature engineering has been used to enhance the detector’s effectiveness.

• We have improved model interpretability and clinician trust by integrating explainable AI techniques, including SHAP and LIME, to explain predictions and validate model decisions.

• We have conducted a comprehensive evaluation against state-of-the-art models, demonstrating that the ETDD system achieves a high accuracy of 96.24%, establishing its effectiveness as a clinical decision-support tool.

Dynamic Modeling of Solar Shading Effects in Photovoltaic Power Generation System

Solar photovoltaic (PV) systems are highly vulnerable to changes in irradiance, and shading is one of the most detrimental issues affecting their efficiency. Even minor shading can cause disproportionate power losses, current mismatch, and the emergence of multiple peaks in I–V and P–V characteristics, complicating maximum power point tracking (MPPT). To address these challenges, this study develops a modeling and simulation framework to systematically analyze shading effects on PV arrays. The model evaluates different irradiance distributions, partial shading scenarios, and the effectiveness of bypass diodes in minimizing mismatch losses. Simulation outcomes demonstrate the direct influence of shading patterns on the I–V response, the reduction in maximum power output, and the nonlinear behavior that hinders MPPT accuracy. Incorporating bypass diodes is shown to mitigate hotspot risks and recover part of the lost power under shaded conditions. Overall, this research highlights the value of simulation tools in predicting real-world PV behavior, offering a practical framework for researchers and system designers to quantify shading impacts and optimize PV performance for enhanced energy yield and reliability.

Autonomous Self-Powered Solar Panel Cleaner Design with Linear Resonant Actuators and Machine Learning

This research presents a novel, waterless, self-sustaining autonomous solar-panel cleaner that combines LDR sensing for dust monitoring, LRA vibration for dust removal, ML-based scheduling, and heat to energy harvesting using TEGs, achieving 95% dust clearance, consuming only 0.366 Wh/day, costing approximately $35 per unit, and delivering an estimated 70% reduction in maintenance costs.

Results of a research on the operation of CoAP, MQTT and 6LoWPAN algorithms in Internet of Things telecommunication networks

Abstract. The article presents the results of a research of the CoAP, MQTT and 6LoWPAN protocols used in Internet of Things (IoT) telecommunications networks. A comparative analysis of the protocols’ performance in transmitting data under resource-constrained conditions typical of IoT environments, such as low bandwidth, high latency, and limited power consumption, is conducted. The research examines the key characteristics and features of each protocol: CoAP implements a lightweight application-layer protocol for exchanging data between devices with limited computing resources; MQTT is focused on transmitting messages in a publish/subscribe mode, ensuring high reliability and scalability of systems, 6LoWPAN provides efficient adaptation of IPv6 to low-data-rate networks. Experimental analysis showed that CoAP exhibits the lowest packet transmission latency for small data volumes, while MQTT performs better with a large number of nodes and high loads. 6LoWPAN, in turn, optimizes the network layer and reduces node power consumption. The obtained results allow us to conclude that the combined use of these protocols is feasible depending on the requirements of specific IoT scenarios—from environmental monitoring to smart home systems and the Industrial Internet of Things.

Calculation of Reliability of Sensors and Detectors in Intelligent System of Telecommunication Network Using Fuzzy Set Methods

The paper presents the calculation of the reliability of sensors and sensors in the intelligent telecommunication network system using fuzzy set methods. The relevance of using fuzzy set methods and fuzzy neural networks in the analysis and study of the main reliability parameters of Internet of Things sensors is shown. It has been shown that neural networks can effectively predict sensor failure, time to failure, or classify the state of a sensor, which increases the reliability of the entire system and allows for timely measures to be taken for its maintenance. Genetic algorithms provide a powerful tool for optimizing mathematical models, especially in problems with many variables and complex dependencies, such as calculating the reliability of sensors. This method allows for efficient search for optimal solutions in multidimensional spaces and can be used to predict failures, estimate time to failure, or other characteristics that affect the reliability of a system. Modeling the reliability parameters of sensors using genetic algorithms allows for efficient optimization of complex models that include many variables and nonlinear dependencies.

Decision-Making Framework for Sustainable Management of the Aral Sea Desiccated Basin under Uncertainty

The desiccated basin of the Aral Sea, once a major inland water body, exemplifies the severe ecological consequences of long-term mismanagement and climate-induced stress. Addressing this environmental crisis requires decision-making frameworks capable of balancing environmental, social, and economic objectives amidst significant uncertainty and data incompleteness. This study proposes a decision-support model grounded in Intuitionistic Fuzzy Set (IFS) theory, integrated within a Multi-Criteria Decision-Making (MCDM) approach, to manage the complex trade-offs inherent in the sustainable restoration of the Aral Sea region. By incorporating hesitation degrees alongside traditional membership and non-membership values, the model captures expert uncertainty and conflicting information more effectively than classical fuzzy or deterministic methods. A real-world case study – focused on water allocation and land rehabilitation – demonstrates the practical utility of the framework. The results highlight the model’s ability to enhance decision robustness, transparency, and reliability in uncertainty-dominated ecological systems, offering a scalable tool for environmental policy and sustainability planning in similar high-risk regions.

Medipath: An Intelligent Emergency Medical Service System with NLP and Real-Time Coordination

This paper discusses the creation and assessment
of MediPath, a smart web-based emergency medical assistance
system. It connects patients, hospitals, and ambulance services
in one digital platform. The system uses technologies like
Natural Language Processing (NLP) to match hospital
specialties with patient symptoms. It also uses real-time location
services and smart route planning to address gaps in emergency
medical response. MediPath has a three-part structure that
supports logins for patients or attendants, hospital management
portals, and navigation systems for ambulance drivers. The
platform uses machine learning to link patient symptoms with
hospital specialties, Firebase Realtime Database for easy data
syncing, and mapping APIs for better traffic-aware routes.
Performance tests show average response times of under 30
seconds for hospital matches and 95% accuracy in matching
symptoms to specialties. There is also a significant drop in
ambulance dispatch delays. User studies at different healthcare
facilities indicate improved coordination, shorter emergency
response times, and better use of resources. Although there are
challenges with highly specialized medical conditions and
connectivity in rural areas, MediPath greatly enhances
emergency medical service delivery by creating a unified
ecosystem that connects all parties in real time. This research
presents a scalable, smart solution that addresses disconnected
emergency medical services. It uses modern web technologies
alongside healthcare knowledge to help save lives through
quicker and more coordinated emergency responses.

Youth Perception of Electric Vehicles: Harmonizing Employment, Environment and Sustainability

understanding urban citizens behavior

A Conceptual Framework for the Buying Intention of Energy Efficient LED Bulbs: Evidence from A Developing Country

Environmental challenges and their risks to human health have drawn the attention of academics, policymakers, and industry leaders. This study examines the factors influencing purchase intentions for energy-efficient LED bulbs in the context of sustainable development, which are relevant from business, marketing, and economic perspectives. Moreover, energy-efficient LED bulbs contribute to the attainment of the Sustainable Development Goals (SDGs) through their direct and indirect impacts. By promoting clean energy and sustainable urban growth, energy-efficient LED bulbs are consistent with SDGs 7, 11, and 13.
Theoretically, the research extends UTAUT2 to explain consumer behavior in the context of green technology adoption. Practically, it provides insights for segmentation, targeting, and brand positioning in emerging markets. Managers and stakeholders are encouraged to focus on product design, packaging, quality, energy efficiency, and cost to enhance adoption. Social factors, including family and peer influence, play a key role, suggesting strategies such as social proof, influencer marketing, and media campaigns to boost engagement and loyalty.

From a societal perspective, promoting energy-efficient LED bulbs supports sustainable consumption, energy savings, and community well-being. Economically, the findings guide producers and marketers in building brand value, competitive pricing, and using digital platforms to educate consumers. At the national level, the study emphasizes government support in regulating fair pricing, ensuring sustainable production, and offering financial incentives such as subsidies, tax breaks, and loans. Policymakers can align environmental objectives with national strategies, foster global collaborations, and promote adoption to reduce carbon emissions.Overall, the study contributes theoretically by enhancing UTAUT2, offers actionable insights for business and policy, and supports Bangladesh’s progress toward sustainable development through widespread adoption of energy-efficient technologies.

A Framework for a Green IoT Blockchain Environment: From the Smart Farming Perspective

We propose an energy-efficient blockchain-IoT framework for smart farming that integrates a lightweight Tangle consensus protocol with renewable energy sources to reduce energy consumption and carbon emissions while maintaining security and scalability.

Correlative Analysis on The Effect of Time and Concentration Variation of Various Enzyme Wash (Acidic, Neutral and Granular) On Denim Garments

Dr. Waziha Farha – Conceptualization, Project Administration, Supervision, Writing – Original Draft, Writing – Review & Editing
Dr. Umme Habiba Rahman – Writing – Review & Editing
Rejaul Karim – Writing – Review & Editing
Md. Alauddin Sany – Data Curation
Md. Limon Hosen Miazi – Data Curation
Sanid Al Rafsan – Data Curation
Tanvir Hossain Ashiq – Data Curation

A Multi-functional Smart Pillbox for Elderly Patients: Integrating Reminders, Health Monitoring, and Emergency Features with AI-based Health Condition Prediction

1) Integrated Hardware and Software Solution: We de-
veloped a comprehensive system combining a smart
medicine reminder box with a mobile application and a
web dashboard, ensuring seamless interaction between
physical medication management and digital health
monitoring.
2) Enhanced Medication Adherence and Emergency Re-
sponse: The system features intelligent reminders
(LEDs, buzzer, app notifications) for timely medication
intake and includes a critical panic button for immediate
emergency alerts via SMS and app notifications to
caregivers
3)

IoT-Based Automated Hydroponic System for Smart Agriculture

This work develops a low-cost, solar-powered IoT hydroponic system integrating multi-sensor monitoring, automated nutrient control, cloud logging, and seamless grid fallback, validated on Nutrient Film Technique (NFT) spinach with stable performance and predictive dataset generation.

Multi-Modal Fusion Tuberculosis Severity Classification: Vision Transformer for Chest X-Ray and Tabular Transformer for Clinical Data

1) A novel multimodal fusion framework for TB severity
classification, integrating chest X-ray images via pre-
trained Vision Transformer (ViT) and clinical data via
hyperparameter-tuned Tabular Transformer, with early
feature concatenation for joint learning.
2) Extensive hyperparameter tuning and regularization
(e.g., dropout, learning rate scheduling) to enhance generalization.
3) Achieved 94.01% accuracy on the MIMIC-CXR dataset
using pneumonia as TB proxy, outperforming unimodal
baselines and state-of-the-art methods.

An Efficient Framework for Suicide Risk Detection Integrating Linguistic and Emotional Features Using Graph Neural Network

A summary of the contributions of our study is as follows:
1. A new graph-based framework leveraging social media data for suicide risk detection is proposed.
2. The integration of semantic (GloVe), syntactic (dependency parsing), and emotional (SenticNet) features in a single graph structure is proposed, correlating different perspectives of the data.
3. Here, we demonstrate the creation of more context-sensitive and interpretable models that would outperform certain recent ML, DL, and LLM approaches.
4. The performance of both GraphSAGE and HGNN was significantly improved when the dataset size was increased from 4K to 8K. GraphSAGE kept its reliability high, while HGNN achieved consistent best scores due to its capacity to scale and hence improve the accuracy, recall, and robustness with larger data.

Wood Detection Method with Machine Learning and Correction by Rule-Based Approach

Proposal of a Correction Algorithm for Undetected Timber Using Machine Learning.
While about 93% of timber are detected by machine learnin, about 98% of timber are detected by the proposed method without degrading the F1 score.

Unraveling the Genetic Tapestry: Ontological Mapping and Survival Modeling of Gene Expression Biomarkers for Autism Spectrum Disorder detection

The following are the primary contributions of this study:
* Proposed a remarkable 2-step feature extraction technique that blends statistics and ML to identify ASD-genetic features.
* Introduced a novel Gene Pathway Analysis framework to identify ASD biomarkers from high-dimensional genomic data.
* Conducted Ontological Analysis to discover critical gene connections, that aids in the exact diagnosis and an awareness of autism.
* Performed Survival Analysis on gene data, uncovering key time-linked biomarkers associated with autism progression.

Latent Representation and Generative Augmentation with Graph-Based Learning for Imbalanced Leukocyte Cytomorphology Classification

In this work, we propose a hybrid framework that integrates autoencoder-based latent representation, GAN-driven minority-class augmentation, and graph-based classification with GraphSAGE. Our key contributions are as follows:
1. We propose a latent-space augmentation strategy that synthesizes minority-class embeddings within the autoencoder feature manifold, avoiding artifacts common in pixel-level augmentation.
2. We design a graph-based relational learning framework that embeds leukocyte representations into a similarity graph and applies inductive classification to leverage contextual neighborhood information.
3. Through extensive experiments on the AML-Cytomorphology-LMU dataset, our method achieves ~91% accuracy and improved macro-F1 scores, particularly enhancing recall for rare subtypes.
4. We show that latent augmentation improves minority-class sensitivity and reduces cross-validation variability, yielding more robust and clinically deployable models.

Real-Time Neural Network Framework for Sub-5 ms Anomaly Detection in NIDS

Network Intrusion Detection Systems( NIDS) are critical for securing modern network infrastructures; still, being deep knowledge- predicated approaches constantly suffer from high conclusion quiescence, limiting their connection in real time and edge computing surroundings. This paper presents an optimized Convolutional Neural Network( CNN)- predicated NIDS frame designed to achievesub- 5 ms anomaly discovery while maintaining high discovery delicacy. The proposed ar chitecture employs depthwise separable complications to reduce computational exodus, INT8 quantization for conclusion accel eration, and channel community to overlap packet internee, preprocessing, and type tasks. Evaluation was conducted on three standard datasets — CICIDS2017, UNSW- NB15, and NSL- KDD — demonstrating delicacy situations above 96 and increment exceeding 15,000 packets per second on an NVIDIA Jetson Xavier NXedge device. The frame achieved up to 60 × hastily conclusion than LSTM- predicated births and demonstrated strong zero day discovery performance, achieving 94.8 recall on unseen Botnet attacks. These results validate the proposed system’s capability to deliver both high- speed and high- delicacy discov ery, making it suitable for deployment in quiescence-sensitive, resource- constrained surroundings. future advancements will explore confederated knowledge for distributed model updates, bettered inimical robustness, and adaptive explainability features to ensure secure decision- making in dynamic network surrounds.

FashFit: Personalized Outfit Illusion Using Face and Garment Matching

The significant research contribution of this work lies in the development of FashFit, a personalized outfit recommendation and visualization system that goes beyond typical e-commerce virtual try-on tools. Instead of focusing on sales, it acts as a personal fashion advisor by analyzing user-specific features such as face shape, body size, skin tone, and fabric preferences, and then matching them with garment attributes like color, silhouette, and fabric type. Using computer vision and image processing, the system generates a virtual illusion of the user wearing the suggested outfits, allowing individuals to visualize suitability before making clothing choices. This contributes to fashion technology by filling the gap between simple visualization and true personalization, helping users make confident outfit decisions and reducing the trial-and-error process in clothing selection.

Biotechnology Inspired 3D Printing Structures for Improved Mechanical Performance
VISION BASED TRAFFIC CONTROL SYSTEM FOR PEDESTRIAN SAFETY

This research presents a vision-based dynamic traffic signal control system that leverages computer vision and deep learning (YOLOv8) to improve pedestrian safety and traffic efficiency. Unlike traditional fixed-time signals, the system dynamically adjusts pedestrian crossing times based on real-time pedestrian density, ensuring safe crossings while minimizing vehicle delays. It further incorporates automated violation detection, including red-light running, jaywalking, and stop-line violations, supported by license plate recognition for enforcement. With high detection accuracy (95.4%) and near real-time response, the proposed system offers a scalable and intelligent solution for modern urban mobility. Its contribution lies in combining adaptive traffic light optimization with automated violation monitoring, laying the groundwork for integration into future smart city infrastructure.

Decentralized Access Control Using Blockchain and Smart Contracts for Enhanced Cybersecurity

This paper identifies and outlines the challenges regarding cybersecurity in centralized access control architectures.
The paper proposes a decentralized access control architecture using smart contracts deployed on a blockchain. This system eliminates centralized points of failure and provides automated, tamper-resistant access decision-making using role-based policies.

Pedestrian Flow Analysis Method in Public Spaces by Integrating Visual Information and Pedestrian Model

This paper proposed a method to improve the accuracy of pedestrian flow analysis by applying machine learning-based object detection to urban video data. To handle occlusions and maintain robust tracking, a pedestrian modeling approach was introduced, which allows detection even when objects overlap and enables class identification using past detection results. The effectiveness of the proposed method was verified in a real-world setting. It was confirmed that the method can achieve stable detection and reduce class assignment errors, achieving a mean absolute percentage error of 1.0% for pedestrians and 2.2% for bikes.

Clegora – Legal ChatBot for Indian Consumer and Civil Rights

The current paper brings a design, development and evaluation of an AI-based legal assistant named Clegora to suit the requirements of the Indian legal system. However, based on a state-of- the-art large language model, Groq Llama 3.2, deployed in a Retrieval-Augmented Generation (RAG) framework, Clegora delivers contextual and high-quality responses through the dynamic retrieval of relevant legal documents in a specialized vector database. The system uses a new token management approach that scales input lengths based on query complexity with the aim of being efficient in resource utilization without compromising on the quality of responses. Clegora has privacy and ethical standards, as it is built with secure authentication of the user by Firebase and strong content safety measures. Performance tests show short response times of less than five seconds to simultaneous users and citation efficiency of more than 95%. The responses of the users point out how the assistant could be used to simplify complex legal cases, multi-turn conversations, and provide users with confidence when handling legal cases. Though the existing constraints are the processing of language diversity in the region and very narrow domain queries, the modular design and continual data refreshing makes Clegora a scalable solution that will democratize the access to legal information, enabling informed decision-making in diverse Indian populations.

EfficientNetV2S with End Ensemble for Robust Bangla Handwritten Character Recognition

Recognition of handwritten characters is an important task in the field of image processing. Bangla handwritten characters exhibit greater complexity and variation in shape, along with high inter-class similarity, making their recognition particularly challenging. To address this, we propose the “EfficientNetV2S” model, designed to recognize complex and visually similar characters in the Bangla language. Although Bangla is one of the most spoken languages in the world, research on Bangla character recognition is still comparatively limited. In this study, we apply a deep learning model trained and tested on a custom dataset of Bangla handwritten characters. The model is utilized both for feature extraction and for its proven efficiency in image classification tasks, enabling it to learn image features effectively and recognize characters with remarkable accuracy. Our approach achieves an impressive 96.63% accuracy, highlighting the strength and reliability of the proposed method. To broaden the range of classes, we combined the “BanglaLekha-Isolated” and “Matrivasa-raw (Ekush)” datasets. The results show that the End Ensemble technique effectively tackles recognition challenges and offers strong potential for real-world applications like automation and education. This research advances Bangla character recognition and encourages further exploration in the field.

IGNN: An Attentional Graph-Based Approach for Vision Transformer-Empowered Plant Disease Detection

This study’s primary contribution is the development of a novel ViT-GNN hybrid deep learning model callled IGNN for plant disease classification. By combining the feature-learning prowess of a Vision Transformer with the relational-modeling capability of a Graph Neural Network, the model achieves superior performance and interpretability over traditional methods. This work demonstrates a new paradigm for leveraging both rich visual features and structural dataset relationships, advancing the field of automated plant disease diagnosis.

Comprehensive Analysis of Machine Learning Models on Physical Layer Anomaly in Smart Grids

It contributes a lot to provide information about selection of best machine learning models for anomaly detection.

Techno-Economic Evaluation of Solid-State and Sodium-Ion Batteries in E-Mobility Using a MATLAB Tool

The selection of suitable battery technologies is a key lever in shaping the global transition to sustainable mobility. This study investigates the techno-economic potential of two emerging battery types – solid-state batteries and sodium-ion batteries. While solid-state batteries promise higher energy densities, sodium-ion batteries offer advantages in material availability and potential cost stability due to the abundance of sodium. A MATLAB-based evaluation is used to conduct a holistic analysis focusing on their application within the Tesla Model Y and the VW ID.3. The assessment considers key indicators such as energy density, specific costs, CO2 emissions, and achievable driving range. The results highlight technological trade-offs and demonstrate how different battery choices impact the ecological and economic performance of electric vehicles. Moreover, the flexibility of the modeling approach enables its application across a range of current and future vehicle configurations, supporting strategic decision making in battery selection.

Advanced Machine Learning Models for Prediction, and Performance Optimization in Renewable Energy

The research by Mohammed Alghassab, conducted at Shaqra University, significantly advances renewable energy systems through the application of advanced machine learning (ML) models—Random Forest, Support Vector Regressor, Gradient Boosting, and CatBoost. Key contributions include achieving high prediction accuracy (Gradient Boosting: 94.2% classification accuracy, 1.86 MW RMSE) and perfect scalability prediction (CatBoost: 1.0 accuracy) for solar, wind, hydro, and geothermal systems. These models enhance energy output forecasting, resource classification, and performance optimization by leveraging feature engineering and hyperparameter tuning. The study demonstrates a 12% accuracy improvement and 10% error reduction over baselines, supporting grid stability, cost-efficiency, and CO2 reduction. By addressing intermittency, scalability, and dependability challenges, the research aligns with global sustainability goals, fostering innovation in smart grids and policy-driven energy planning for a low-carbon future.

Data Resilience in Cyber-Physical Systems: Overcoming MQTT Data Loss for Reliable AI Subsystems

The paper introduces and experimentally validates an end-to-end data-resilience layer for MQTT-based CPS that maintains AI subsystem reliability under adverse network conditions. Concretely, it contributes practical mechanisms (loss-aware buffering/retransmission, QoS/retention strategies, deduplication/back-pressure, and gap-tolerant ingestion for AI) and shows, through controlled tests with packet loss and instability, that operational continuity and model performance can be sustained despite communication faults.

AI Impact on Sustainable Economy Growth

The paper proposes an integrated Green Finance–AI framework and platform that links financial allocation with AI-enabled monitoring, verification, and reporting. This framework allows banks to transition from symbolic commitments to measurable decarbonization strategies. By aligning green finance with AI capabilities, the proposed solution supports environmental protection, social inclusion, economic resilience, governance accountability, and digital innovation; thereby operationalizing sustainable economic growth across all five pillars

Market-Based Pricing Scheme for Isolated Microgrids with Renewable Energy and Storage

An integrated, market-based pricing framework for isolated microgrids that jointly optimizes hourly PV–wind–battery–diesel dispatch and derives LCOE-consistent tariffs

Quantum Threat Defense: A Framework for Migrating IoT-Based Healthcare Systems

This study is significant because it directly addresses the emerging quantum threat to IoT-based healthcare systems, which represent one of the most critical and vulnerable sectors of modern infrastructure. While much of the existing literature explores general post-quantum cryptography (PQC) migration strategies, there is a lack of frameworks specifically tailored to the unique constraints of healthcare IoT ecosystems. By proposing a phased, hybrid migration framework that accounts for the layered IoT architecture, device heterogeneity, and resource limitations, this paper provides a practical and structured pathway for ensuring security in healthcare systems against future quantum-enabled attacks.

The framework’s emphasis on crypto-agility, interoperability, and phased adoption is particularly important for healthcare environments where continuous operation and patient safety are paramount. Furthermore, the study bridges a critical gap by integrating both technical and operational perspectives, making it valuable for policymakers, system architects, and healthcare providers preparing for the quantum era. Ultimately, this work contributes to safeguarding patient data, ensuring the reliability of medical devices, and preserving trust in digital healthcare services in the face of advancing quantum computing capabilities.

Medipath: Intelligent Emergency Health Navigator

This paper discusses the creation and assessment
of MediPath, a smart web-based emergency medical assistance
system. It connects patients, hospitals, and ambulance services
in one digital platform. The system uses technologies like
Natural Language Processing (NLP) to match hospital
specialties with patient symptoms. It also uses real-time location
services and smart route planning to address gaps in emergency
medical response. MediPath has a three-part structure that
supports logins for patients or attendants, hospital management
portals, and navigation systems for ambulance drivers. The
platform uses machine learning to link patient symptoms with
hospital specialties, Firebase Realtime Database for easy data
syncing, and mapping APIs for better traffic-aware routes.
Performance tests show average response times of under 30
seconds for hospital matches and 95% accuracy in matching
symptoms to specialties. There is also a significant drop in
ambulance dispatch delays. User studies at different healthcare
facilities indicate improved coordination, shorter emergency
response times, and better use of resources. Although there are
challenges with highly specialized medical conditions and
connectivity in rural areas, MediPath greatly enhances
emergency medical service delivery by creating a unified
ecosystem that connects all parties in real time. This research
presents a scalable, smart solution that addresses disconnected
emergency medical services. It uses modern web technologies
alongside healthcare knowledge to help save lives through
quicker and more coordinated emergency responses.

FedFall: A Federated Transfer Learning Framework for Privacy-Aware Fall Detection in Elderly Care Using IoT-Edge Devices

1) FedFall framework: a novel architecture that fuses federated and transfer learning to enable privacy-preserving and personalized fall detection.
2) Edge-based deployment strategy: an end-to-end system
that supports real-time inference and training on low-
power IoT devices.
3) Simulation-based validation: demonstration of FedFall’s feasibility using simulated multi-client datasets, high lighting accuracy, latency, and communication efficiency.
4) Adaptation for constrained devices: lightweight models and compression techniques to ensure compatibility with embedded platforms.

Fast and Memory-Efficient 4SQR-Code Decoding using Histogram Analysis

Firstly, we introduce a histogram-based decoding method for 4SQR-Code that improves efficiency over traditional approaches.

Secondly, we demonstrate substantial improvements in
both decoding time and memory usage on 3,000 test
samples, achieving an average reduction of approximately 98.66% in decoding time and nearly 88.91% in memory consumption compared to baseline methods.

Finally, we conduct a comparative analysis with the
traditional OpenCV QR Code detector to demonstrate the advantages of our method.

Stability Enhancement and Harmonic Suppression of a Five-Level Cascaded H-Bridge Inverter for Microgrid Systems using Optimized PI Control

Integrating renewable energy into microgrids requires advanced power conversion systems that ensure stability, power quality, and harmonic suppression. This paper presents the design, analysis, and optimization of a five-level cascaded H-bridge (CHB) inverter integrated with an LCL filter and controlled by a PI controller using sinusoidal pulse-width modulation (SPWM). This work focuses on enhancing stability and harmonic performance in sustainable energy microgrids through parametric tuning of PI controller gains (Kp, Ki) and comparative analysis with alternative control strategies. Simulation results demonstrate that optimal PI tuning achieves a total harmonic distortion (THD) of 2.8% in grid current, with a robust transient response under dynamic loads. The sinusoidal pulse width modulation (SPWM) technique offers straightforward implementation while supporting compatibility with multilevel inverter topologies, making it scalable for high-voltage applications. Comparative analysis with proportional-resonance (PR) and hysteresis controllers emphasizes the practicality of PI controllers, achieving an effective balance between computational simplicity and control performance. Our proposed system addresses the main challenges of integrating renewable energy and provides a better and more cost-efficient way to enhance microgrid resilience and power quality. Future research should focus on more reliable and hardware-based authentication to transition key findings from the simulation into real-world applications

Experimental Investigation of the Impact of Partial Shading Size on Power Losses in Outdoor Solar Modules: Insights from Five Case Studies

Partial shading significantly affects the performance of photovoltaic (PV) systems, resulting in significant power losses and decreased efficiency. This study examines the effect of partial shading on the output power and electrical parameters of photovoltaic modules through field case studies and outdoor experimental tests. Additionally, an I-V tracer, PROVA 1011, was used to experimentally examine 10W and 20W polycrystalline panels under controlled partial shade sizes of 20%, 40%, 50%, 60%, and 80%. The results show that the voltage remains stable at lower shading levels but drops sharply beyond 50%. The current declines linearly with increasing shading, and the power losses reach 88.4% under 80% shading. Five case studies have been considered in this study to analyze rooftop-based partial shading size and the loss calculated using the experimental data. The study shows that power losses vary from 83.87 % to 87.91% due to the partial shading size of 40% to 60% for a 100W module. These findings emphasize the critical impact of rooftop obstacle-based partial shading on PV system losses and the need for practical design of PV installation to mitigate the shading.

Impact of Seasonal Climate Variations on EV Battery Performance and Charging Efficiency in Dhaka, Bangladesh

Electric vehicles (EVs) are central to sustainable transportation; however, their battery lifespan and charging efficiency are significantly impacted by seasonal temperature and humidity variations. This study investigates how Dhaka’s distinct climate periods—winter, summer, monsoon, and post-monsoon—influence EV battery degradation and charging performance. Laboratory tests at 10 °C, 25 °C, and 40 °C quantify capacity fade and impedance growth over 1,000 cycles, while field-deployed battery packs record real-world temperature and humidity impacts each season. Charging behavior is evaluated using 7 kW AC Level 2- and 50-kW DC fast chargers, measuring charge time and efficiency under ambient conditions. Results show that sustained summer heat (≥ 35 °C) accelerates irreversible capacity loss by approximately 15% per 1,000 cycles, and monsoon humidity (≥ 85% RH) reduces charging efficiency by about 8%. Recommendations include adaptive liquid-cooled battery thermal management, ambient-aware charger derating, and humidity-resistant connector designs. These strategies aim to extend battery life, improve charging reliability, and support EV adoption in tropical, monsoon-prone environments.

A Unique Design of a Multi-Axial Movable Bed for Residential Use by Bedridden Individuals

1. By overcoming current limitations like production cost, operational technology, and non-availability of advanced technology, this work designed a multi-axial movable bed system for bedridden individuals.

2. This design entails the integration of locally sourced components and parts, and real-time control of multidirectional kinematics.

3. The proposed design aspires to enhance patient autonomy, reduce caregiver workload, and provide a scalable, sustainable assistive solution tailored to the biomechanical and socio-technical requirements of domiciliary care in resource-constrained environments.

Explainable Hybrid Intrusion Detection: Combining Ensemble Learning and XAI for Transparent Cybersecurity

• The developed an advanced intrusion detection system
achieved 0.97 accuracy and 0.99 precision for attack clas-
sification, significantly outperforming baseline models
• This study innovatively integrates SHAP and LIME in-
terpretability methods, providing both global and local
explanations that validate the model’s reliance on mean-
ingful network features such as packet time-to-live (sttl)
and protocol types (proto).
• Through rigorous evaluation metrics (AUC=0.9974) and
error analysis , the study establishes the reliability of
the model for real-world deployment in security-sensitive
environments.
• This study contributes to cybersecurity research by
demonstrating how interpretable machine learning can
effectively detect modern network threats while maintain-
ing operational transparency for security analysts.
• The study outlines critical pathways for subsequent work,
including adaptive learning mechanisms for evolving
threats and optimization of edge computing architectures
in distributed networks

A Multi-Strategy Ensemble Learning Framework for Robust and Scalable Malware Detection in Cybersecurity

1.A universal ensemble system with high performance levels was developed according to metric measures and compared to the baseline ML and DL models on the EMBER dataset.
2.Designed an integrated system that uses RandomForest, Extra Trees, XGBoost, LightGBM, Hist-GradientBoosting and unified deep-learning ensembles to control bias-variance in a balanced way.
3. Both traditional and deep learning models incorporate soft-voting, stacking, and meta-learning to ensure adaptability across diverse malware feature spaces.
4. Demonstrated scalability using memory-mapped data handling and standardized preprocessing for large datasets

Between Innovation and Oversight: A Cross-Regional Study of AI Risk Management Frameworks in the EU, U.S., UK, and China

Providing the first systematic comparative analysis of the practical trade-offs between the EU’s rights-based, the U.S.’s market-driven, the UK’s flexible, and China’s state-controlled AI governance models, offering a crucial framework for global policymakers.

Enhancing Team Collaboration through Social Network Analysis: A Transactive Memory System-Based Visualization Framework

This research develops a Transactive Memory System-based visualization framework that leverages Social Network Analysis to enhance team collaboration in project management. The contribution lies in integrating SNA metrics with an interactive dashboard to provide managers with actionable insights into communication structures and knowledge sharing. The framework was validated through pre- and post-analysis in real organizational settings, demonstrating its potential to improve coordination, decision-making, and overall project performance.

BanglaMediText: A Comprehensive Resource for Bengali Medical Text Classification with Classical, Neural, and Transformer Approaches

Md. Shohanur Rahman Shohan: Data analysis, Software, Methodology, Validation, Writing – original draft. Md Habibur Rahman: Conceptualization, Methodology, Software, Validation, Writing – reviewing, editing, and Supervision. Md Abbas Mahmud Suzon: Data curation, Validation, Writing –review and editing. Md Imran Hasan: helped in the preparation of the tables and figures. Md Shofiqul Islam: Review and Editing, Md Aktaruzzaman: Review and Editing.

Benchmarking Bengali Dialect Identification: A Comparative Study of Machine Learning, DNN, and Transformer Architectures

Md. Shohanur Rahman Shohan: Data analysis, Software, Methodology, Validation, Writing – original draft. Md Habibur Rahman: Conceptualization, Methodology, Software, Validation, Writing – reviewing, editing, and Supervision. Md Abbas Mahmud Suzon: Data curation, Validation, Writing –review and editing. Sabrina Ferdous: helped in the preparation of the tables and figures. Md Shofiqul Islam: Review and Editing, Md Aktaruzzaman: Review and Editing.

Optimization and Performance Evaluation of Semi-Transparent Photovoltaic Panels for Sustainable Urban Energy Solutions

Solar energy integration in the future of rooftops, windows, and interiors of glass as urban areas grow, and the need to provide power rises. Building-integrated photovoltaics (BIPV) is a new concept, where semi-transparent solar cells can turn glass facades into power plants without compromising comfort, daylight, or styling. This recent generation of solar technological solutions can solve the problem of sustainable energy production. The current work investigates the impact of transparency levels on energy generation by modeling the optical and electrical responses of ST-PSCs using finite-difference time-domain (FDTD) and SCAPs-1D software. Additionally, the study is devoted to performance assessment in the building-integrated environment and offers an optimized design architecture to be implemented in practice. The simulation outcomes show that decent efficiencies of 16.41% can be acquired with moderate transparency of 31.7% which sheds light on the significance of ST-PSCs in vertical and urban designs. The paper uses cell design software to simulate and optimize these semi-transparent panels. The findings highlight the feasibility of using semi-transparent solar panels as an optimal solution that addresses land scarcity, energy efficiency, and urban energy needs, providing valuable insights for the future integration of renewable energy systems in urban infrastructure and agriculture.

A study on the construction of the simulated patient agent using non-verbal information in Objective Structured Clinical Examination

This study proposes and verifies a method to estimate speech endings in medical interviews using non-verbal information like face direction and tone. This could lead to the development of more realistic virtual patients for Objective Structured Clinical Examination (OSCE), addressing the shortage of human simulated patients and improving training for medical students.

MLA-Net: Attention-Driven MobileNet–LSTM Framework for Robust Alzheimer’s Disease Detection from MRI Data

Md Abbas Mahmud Suzon: Data analysis, Software, Methodology, Validation, Writing – original draft. Md Habibur Rahman: Conceptualization, Methodology, Software, Validation, Writing – reviewing, editing, and Supervision. Md Imran Hasan: Data curation, Validation, Writing –review and editing. Md. Shohanur Rahman Shohan: helped in the preparation of the tables and figures. Md. Robiul Hoque: Review and Editing, Md Aktaruzzaman: Review and Editing, Md. Farukuzzaman Khan: Review and Editing.

Multilocation sequential navigation for autonomous mobile robots based on IR-UWB ranging

Traditional AMR navigation methods require prior mapping or the setup of a coordinate system for interested space, which can impose significant overhead.
We propose and implement an autonomous mobile robot (AMR) navigation system based on impulse radio ultra-wideband (IR-UWB), which relies solely on IR-UWB ranging between IR-UWB terminals mounted on the AMR and those placed at designated locations. Additionally, we incorporate transmit power control (TPC) to mitigate multipath effects and apply a moving average filter to smooth route fluctuation. We implement the navigation system using this approach and validate its effectiveness through experimental measurements.

Knowledge-Defined Networking-Enabled IoT Communication Framework for Resilient Smart Grid Protection

The paper proposes a KDN-enabled IoT framework for resilient smart grid protection, integrating AI/ML-based anomaly detection, predictive fault recovery, and adaptive routing within a three-plane architecture. Simulations and testbed results show notable gains over baseline SDN—25.2% lower latency, 4.5% higher PDR, 43.8% faster recovery, 7.7% better attack detection, and 15.6% lower energy use—demonstrating a scalable, secure, and efficient smart grid communication solution.

Knowledge-Defined Networking-Enabled IoT Communication Framework for Resilient Smart Grid Protection

The paper proposes a KDN-enabled IoT framework for resilient smart grid protection, integrating AI/ML-based anomaly detection, predictive fault recovery, and adaptive routing within a three-plane architecture. Simulations and testbed results show notable gains over baseline SDN—25.2% lower latency, 4.5% higher PDR, 43.8% faster recovery, 7.7% better attack detection, and 15.6% lower energy use—demonstrating a scalable, secure, and efficient smart grid communication solution.

Mixed-Integer Linear Programming for Supply Chain Network Design, Evaluating Upstream–Downstream Sustainability Impacts

A key contribution of this study is the formulation of a holistic mathematical model that simultaneously considers upstream sites selection, transportation routing, and vehicle choice, and evaluates their combined effects on downstream sustainability outcomes. The model employs a general sustainability index that can captures the specific sustainability priorities of the enterprise.

Seed Nano-Priming with Silicon-Rich Silica Nanoparticles: A Novel Approach Enhancing Maize Germination and Early Growth

* Synthesis of Silicon-rich-silica from local river sand using a low-cost process
* Enhancement of seed germination and seedling growth parameters by priming Maize seed with Silicon-rich-silica.