Authors - Jeba Priya J, N. Priya Abstract - Mental health challenges among young adults require innovative psychoeducational interventions. This study presents the development and preliminary evaluation of Dear Alfred, a serious virtual reality (VR) game designed to enhance emotional self-regulation and intergenerational empathy. Grounded in the Process Model of Emotion Regulation, the game immerses players in a narrative- driven experience addressing elderly isolation. The development followed an iterative methodology, resulting in a playable vertical slice tested on Meta Quest 2 and 3 platforms. This work contributes to the field by proposing a scalable, multidimensional approach at the intersection of psychology, technology, and education, highlighting the specific need for hardware-specific optimization in digital mental health solutions.
Authors - Ashwini V. Zadgaonkar, Sonali Potdar, Archana Bopche, Pranali Pawar, Rupali Vairagade, Yogita Hande Abstract - Time series prediction plays a critical role in monitoring and control of electrical power systems, particularly for detecting frequency fluctuations caused by imbalances between generation and demand. This study proposes an early warning framework for frequency fluctuation events using a hybrid k-Nearest Neighbour (KNN) and Dynamic Time Warping (DTW) approach combined with a global confidence interval based decision mechanism. Electricity frequency data collected from the New Zealand power grid over a six-month period were segmented into training, validation, and testing sequences. Alignment distances between historical and incoming sequences were used to identify precursor patterns indicative of impending frequency disturbances. Experimental results show that the proposed method achieves high warning accuracy with a very low false negative rate, outperforming baseline models such as ARIMA and LSTM. The findings demonstrate that KNN–DTW provides an effective and practical solution for early warning of frequency fluctuations, supporting improved operational reliability in modern power systems.
Authors - Arianna Cobb, Vishnu Kumar Abstract - Teen suicide remains a significant public health concern in the Unit ed States, with substantial geographic variation across counties. Understanding how socio-environmental and healthcare access factors relate to suicide risk can help identify communities that may benefit from targeted interventions. This study aims to support this effort by analyzing county-level teen suicide patterns using K-means clustering, an unsupervised machine learning technique. A da taset of 248 U.S. counties with reported teen suicide data was constructed using five-year aggregated suicide crude rates (2019-2023) alongside multiple socio environmental and healthcare indicators, including hospitalization rates, mental health provider availability, primary care provider rates, social association rates, uninsured population percentages, poverty levels, food insecurity, and rural population share. K-means clustering was then applied to identify county-level risk profiles. The results reveal two distinct county groups: one characterized by lower suicide rates, greater healthcare provider availability, stronger social as sociations, and lower socioeconomic disadvantage; and another characterized by higher suicide rates, reduced healthcare access, higher poverty and food in security, and greater rural residency. These findings highlight meaningful coun ty-level disparities and demonstrate the utility of machine learning approaches to identify regional risk profiles associated with teen suicide. The results may help inform public health strategies and policy efforts aimed at prioritizing re sources and expanding mental health services in high-risk communities.
Authors - Adhi Sree Praveen Pai, Alaganandha Pradeep, Jeremy Simon Moncey, Josin Kurian Athikalam, Lakshmi K.S. Abstract - This research investigates the digital footprint of mental health infor mation as it circulates on YouTube. Using a qualitative content analysis ap proach, the study examines 100 selected videos in conjunction with social media analytics to identify recurring patterns in the dissemination of mental health dis course. The findings reveal a mix of misleading or incomplete claims, educa tional resources, personal narratives, and recovery-oriented content, illustrating how mental health discussions shape and amplify user perspectives at both broad (macro) and specific (micro) levels within the evolving field of e-health. To in terpret these dynamics, the analysis applies Gibson’s theory of transactional af fordances, which illuminates key themes of risk, relevance, lived experience, credibility, and social support. By situating these themes within the broader con text of video-sharing platforms, the study underscores the importance of YouTube as a platform for mental health communication. It underscores its role in broader public conversations about health in the digital age. The future re search should investigate mental health discourse from other social media users.
Authors - Rahul Singh, Sachin B. Jadhav Abstract - Cloud cover generally limits the applicability of optical remote sensing images for tasks such as agriculture monitoring and disaster relief. Cloud removal is an inherently difficult problem because of the lack of spatial structures and spectral information. To effectively remove cloud contamination from SAR and optical images, we propose a speckle-aware global cross-attention network. The proposed SAR-optical cloud removal network architecture consists of a dual encoder with a global cross-attention mechanism that allows for effective cross-modal interactions. Additionally, a refining module and symmetric decoders improve the accuracy of the reconstructed image. Furthermore, we propose a speckle-aware gating mechanism to perform speckle filter adaptation. The experimental results affirm that our proposed network outperformed the baseline by increasing Peak Signal-to-Noise Ratio(PSNR) by +0.86 dB, Structural Similarity Index Measure(SSIM) by +0.142, and reducing the spectral distortion of the image. Additionally, we noticed a decrease in the Root Mean Square Error(RMSE) and Spectral Angle Mapper(SAM) values. This infers that selective SAR-Optical fusion with an adaptive noise-aware gating mechanism improves the accuracy of cloud-free optical images and optical remote sensing images.
Authors - Nyuti Bhesania, Khushi Solanki, Bimal Patel, Purvi Prajapati, Priyanka Patel Abstract - The rapid advancement of information and communication technology (ICT) has accelerated the digital transformation of public sector governance, including tax administration. This study examines the impact of Indonesia’s Core Tax Administration System (Coretax) on micro, small, and medium enterprise (MSME) tax compliance within an ICT–behavioral framework. Using survey data from 300 MSME taxpayers and Structural Equation Modeling–Partial Least Squares (SEM-PLS), the study analyzes the direct and indirect effects of Coretax utilization on tax compliance through administrative efficiency and trust in the tax authority. The results indicate that Coretax utilization has a positive and significant effect on administrative efficiency, trust in the tax authority, and MSME tax compliance. Administrative efficiency and trust also significantly influence compliance, con-firming their mediating roles. These findings demonstrate that digital tax administration functions not only as a technological reform but also as an institutional and behavioral mechanism that reduces compliance burdens and strengthens vol-untary compliance. From a sustainable development perspective, improved MSME tax compliance supports Sustainable Development Goal (SDG) 8 by enhancing domestic revenue mobilization for inclusive economic growth, while the integrative and trust-building role of Coretax reflects SDG 17 through strengthened partnerships among government, technology providers, and taxpayers. This study contributes empirical evidence on digital tax systems in developing economies.
Authors - Ajidhashini Thulasidass, M. Suresh Abstract - Modern railway system increasingly rely on digital technologies such as Communication-Based Train Control (CBTC), European Train Control System (ETCS) and Supervisory Control and Data Acquisition (SCADA) systems, raising significant cyber-security challenges. We have seen 220% increase in attacks over five years from opportunistic ransomware to sophisticated targeted threats. This paper provides an overview of railway cybersecurity and surveys the coverage area considering ICT architectures, cyber threat models, and AI-based defense approaches. 75% of cases employed Distributed Denial of Service (DDoS) tactics while ransomware had affected 54% of the OT environments. We describe a comparative taxonomy of Artificial Intelligence and Ma-chine Learning approaches including the methods based on supervised learning, unsupervised learning, and advanced deep learning practices with detection accuracy as high as 97.46%. However, there exist several challenges: few available public data sets, lack of validation in real-world scenarios, demands for explain ability from that AI system and worries about adversarial robustness. We discuss eight potential research gaps, and future directions focusing on federated learning, digital twin development, multimodal AI fusion and safety-security co-engineering frameworks.
Authors - Vishnu Kumar, Natalia Miranda Abstract - Food insecurity remains a pressing public health and equity challenge in urban U.S. communities, with the Supplemental Nutrition Assistance Program (SNAP) serving as the primary federal mechanism for alleviating household food hardship. Despite its importance, SNAP participation varies substantially across neighborhoods, reflecting underlying socioeconomic disparities. This study leverages neighborhood-level data from Baltimore City to identify the key socioeconomic drivers of SNAP participation using explainable machine learning (ML) techniques. Three supervised ML models: Decision Tree, Random Forest, and XGBoost were developed and evaluated using standard regression metrics. The Random Forest model demonstrated the strongest predictive performance. Model interpretability was enhanced through Shapley Additive Explanations (SHAP), which quantified the contribution of each feature to predicted SNAP participation. Results indicate that lower income, shorter life expectancy, higher Temporary Assistance for Needy Families (TANF) participation, higher proportions of female-headed households, and lower educational attainment are associated with increased SNAP reliance. These findings highlight the complex interplay be-tween economic deprivation, social vulnerability, and neighborhood-level assistance utilization, offering actionable insights for policymakers and public health practitioners. By combining predictive accuracy with interpretability, explainable ML provides a robust framework for informing evidence-based interventions aimed at reducing food insecurity and promoting equity in urban communities.
Authors - Hector Rafael Morano Okuno Abstract - This work proposes an intelligent system for automatic food-image-based recognition and calorie estimation to meet the emerging demand for accurate dietary monitoring and personalized nutrition recommendations. Conventional food-logging methods are cumbersome, prone to errors, and mostly fail to capture portion sizes, hence motivating an end-to-end computer vision and depth-based approach. The proposed system utilizes a custom-curated Indian food image dataset of eighty classes, collected, labeled, and preprocessed to make it robust enough to present various variations in lighting, background, etc. A deep learning model was then trained for detecting and classifying food with high precision. The overall classification accuracy achieved by the proposed system is ninety-seven percent. The depth understanding of the detected food regions will provide an approximation of volume and weight, leading to relatively better calorie calculations. Nutritional analysis gets integrated into the system by relating the type and estimated weight of food to the standard nutritional information for detailed insights in terms of calories, proteins, fats, car-bohydrates, fiber, and micronutrient content. The results for evaluation reveal strong detection, minimum estimation error, and efficient real-time processing, which clearly show its applications. In this paper, an approach that combines recognition by image, depth estimation by portion, and nutrition logic capable of leading to a strong solution for diet determination has been introduced.
Authors - D.Swetha, Senthilkumar Selvaraj, K.M.Madhan Prasanth, D.Nihal Abstract - The rapid expansion of digital commerce platforms has significantly transformed on- line transactional systems; however, conventional centralized architectures continue to face critical challenges related to security, transparency, data integrity, and trust management. Traditional e-commerce systems rely heavily on centralized databases, making them vulnerable to data tam- pering, unauthorized access, fraudulent transactions, and single points of failure. To address these limitations, this paper proposes a secure, scalable, and modular web-based e-commerce system that is architecturally designed for integration with blockchain technology and smart contracts. The proposed system is implemented using widely adopted web technologies, with a responsive frontend and a robust backend to support essential functionalities such as user authentication, product catalog management, shopping cart operations, order processing, inventory management, and administrative control. The architecture emphasizes separation of concerns, enabling flexibility, maintainability, and future extensibility. A key contribution of this work lies in the incorporation of a blockchain-ready framework that enables immutable transaction recording and enhanced trace- ability across the entire transaction lifecycle. Smart contracts automate transaction validation and order execution. The system also introduces an AI-based anomaly detection mechanism using a Deep Q-Network to detect fraudulent behavior. Experimental validation demonstrates reliable per- formance and scalability.
Authors - Areej Almazroa, Sara Albahlal, Dalia Alswailem, Dhay Altamimi, Aljoharah Aldaej, Heba Kurdi Abstract - Monitoring marine litter is essential for planetary and human survival. This study proposes a novel framework integrating satellite data and big data analytics to assess marine litter distribution in coastal and oceanic environments. Leveraging open-source imagery from COPERNICUS Sentinel-2 and LANDSAT, the framework utilizes reflectance methodologies and image processing to identify and classify marine debris, focusing on spectral bands from visible blue (490 nm) to short-wave infrared (1610 nm). A pilot case study in San Diego, California, demonstrates the approach’s feasibility. The study explores the potential of microwave radiometry and machine learning for material detection and contour analysis, showing how satellite data can support dynamic and cross-platform monitoring systems. Results validate the use of remote sensing technologies to map plastic debris, providing a replicable methodology that combines emergent (e.g., satellites, drones) and traditional (e.g., sampling) techniques. This approach contributes to a deeper understanding of plastic pollution pathways, sources, and impacts across economic sectors. By generating harmonized data on mismanaged plastic waste, the study informs sustainability strategies and circular economy practices, helping redesign systemic plastic management and supporting local and global environmental governance.
Authors - Sonali S. Gaikwad, Jyotsna S. Gaikwad Abstract - In this semi-systematic literature review, a detailed study of the role of Human-Computer Interaction (HCI) in creating game-based solutions for Attention-Deficit/Hyperactivity Disorder (ADHD) among children is conducted. Six peer-reviewed research studies were selected. The study demonstrates that HCI can serve as a major therapeutic mechanism by transforming digital platform-based cognitive training into engaging, interactive experiences. These approaches not only improve focus but also enhance the overall effectiveness of interventions. Key findings from the analyzed studies are discussed, and future research directions are proposed, including multimodal hybrid systems with adaptive personalization and accessibility features to further improve outcomes for children with ADHD.
Authors - Bharathi A, Mohan Kumar P, Subha B Abstract - Rupture of an intracranial aneurysm results in catastrophic subarachnoid hemorrhage with a 30–40% fatality rate. Although treatment decisions are guided by clinical risk scores (PHASES, ELAPSS), recent research suggests that morphological analysis and computational fluid dynamics (CFD) may offer better rupture prediction. This study looked at 92 middle cerebral artery aneurysms from the CMHA dataset, which included 71 that had ruptured and 21 that had not. We evaluated four feature sets: Clinical-Basic (13 variables), Clinical-Scores (adding PHASES and ELAPSS; 15 variables), Scores and Morphology (24 variables), and Full (28 variables). We trained logistic regression models using 5- fold cross-validation with a 20% test set. We used bootstrap validation (1000 iterations) and Bonferroni-corrected feature importance analysis to reduce overfitting. The AUC for the Clinical-Basic set was 0.891±0.063. Performance was enhanced to a maximum AUC of 0.976±0.034 by adding PHASES and ELAPSS. The Full model achieved an AUC of 0.981±0.029, with neither morphological nor hemodynamic variables giving much further improvement. Significant variance was revealed by bootstrap analysis (95% CI: 0.764-0.998). At 90% specificity, the test set's AUC was 0.933, but its sensitivity was only 14.3%. The primary contributors were ELAPSS (F=143.2, p<10⁻¹) and PHASES (F=38.4, p<10⁻¹), whereas morphological and hemodynamic characteristics did not exhibit any significant correlations. Clinical scores demonstrated strong discrimination, but CFD-derived parameters offered minimal additional value in this small, imbalanced, single-center group. The wide confidence intervals and class imbalance limit clinical recommendations. Further validation in larger, multicenter studies is necessary.
Authors - Tirupathi Rao Dockara, Manisha Malhotra Abstract - AI and data platforms are increasingly expected to deliver end-to-end business automation under rapid market and regulatory change. However, prevailing platform construction strategies remain predominantly top-down: teams standardize a generic capability stack and subsequently customize it for heterogeneous domains through code, integration glue, and service forks. This approach amplifies technical debt, fragments governance, and makes continuous adaptation expensive. This paper introduces the Inverse Vertex Pyramid (IVP), a design pattern that reverses the direction of platform derivation. IVP begins at the use-case vertex by conducting rigorous analysis of high-value specialized automation scenarios and generalizes them into explicit, machine-actionable platform descriptors (metadata models, domain ontologies, policy/workflow specifications, and capability contracts) that form a stable, reusable core. Specialization is realized primarily via declarative configuration and policy changes, rather than code rewrites. We formalize IVP as a pattern, propose a reference architecture separating control and execution planes, and provide a comparative analysis against layered architectures, domain-driven design, and microservice platforms. A proof-of-concept walkthrough in regulated claims automation illustrates the generalization mechanism and highlights how IVP can reduce re-engineering, improve governance consistency, and accelerate time-to-market. The paper concludes with limitations, threats to validity, and a research agenda for automated use-case mining, formal verification of policies, and quantitative evaluation of platform agility.
Authors - Nishant Shah, Ansh Bajpai, Shrivaths S. Nair, Manas Verma K, Sabitha S Abstract - Digital accessibility in higher education is a key requirement to ensure the inclusion of students with hearing disabilities. However, institutional plat-forms often present barriers that limit autonomy, understanding of information, and full participation. The objective of this study was to evaluate the user experience of students with hearing disabilities on the EVIRTUAL, SGA, and SIS platforms of the Technical University of Manabí, identifying perceptions, accessibility barriers, and improvement proposals. A descriptive, exploratory study with a mixed-methods approach was conducted. The population consisted of seventy-eight students with hearing disabilities registered in the Inclusion Unit, from which an intentional subsample of ten participants was selected. A structured sur-vey with Likert-type scales and a participatory observation form were applied in real interaction situations with the platforms. Quantitative analysis was carried out using descriptive statistics, while qualitative information was organized into thematic categories. The results show that half of the participants achieve full autonomy in the use of the platforms, forty percent require intermittent support, and the rest need constant assistance. Regarding clarity of information and con-tent comprehension, intermediate responses predominate, which reveals recur-rent difficulties. The main barriers identified were a confusing interface, non-intuitive navigation, insufficient visual supports, and the need for external assistance. The study proposes improvements such as customizable subtitles, step-by-step visual guides, an accessibility button, a sign language interpreter avatar, and optimization for mobile devices, aimed at strengthening autonomy and user experience.
Authors - Sabarishwaran V, Gomathi K, Andey Phani Vinay, Jagadeeswaran V, Ranjith Kumar M Abstract - The rapid expansion of digital commerce platforms has significantly transformed on- line transactional systems; however, conventional centralized architectures continue to face critical challenges related to security, transparency, data integrity, and trust management. Traditional e-commerce systems rely heavily on centralized databases, making them vulnerable to data tam- pering, unauthorized access, fraudulent transactions, and single points of failure. To address these limitations, this paper proposes a secure, scalable, and modular web-based e-commerce system that is architecturally designed for integration with blockchain technology and smart contracts. The proposed system is implemented using widely adopted web technologies, with a responsive frontend and a robust backend to support essential functionalities such as user authentication, product catalog management, shopping cart operations, order processing, inventory management, and administrative control. The architecture emphasizes separation of concerns, enabling flexibility, maintainability, and future extensibility. A key contribution of this work lies in the incorporation of a blockchain-ready framework that enables immutable transaction recording and enhanced trace- ability across the entire transaction lifecycle. Smart contracts automate transaction validation and order execution. The system also introduces an AI-based anomaly detection mechanism using a Deep Q-Network to detect fraudulent behavior. Experimental validation demonstrates reliable per- formance and scalability.
Authors - Sowmyashree N, Madhu Sunkanur, Impana M, Suchithra B S, Hemalatha P G Abstract - The existence of a growing social media has created complex cyber systems in which vast quantities of interactions constitute substantial issues regarding misinformation, privacy invasion, deception of identities, and destructive behavioural tendencies. The regularity of involvement in this type of big systems requires sophisticated systems that are able to judge the motive of the user, content validity and suspicious activities within real time. Overall interest will be to develop a universal trust calculation system that will be more secure and effective in ensuring privacy and increasing the accuracy of suspicious or malicious users in social sites. The proposed Multi-Layer Federated Trust Framework algorithm is a combination of peer-based user reputation scoring, feature-based content authenticity detection, federated trust indicators aggregation, and anomaly detection with the help of behavioural anomalies. These approaches cooperate with secure aggregation and decentralized learning in removing the uncoded information exposure and enable the computation of trust at scale. The proposed algorithm is experimentally confirmed, and the obtained results are 95.2, 94.1, 93.5, and 93.8, corresponding to a minimum latency of 65 ms and a privacy preservation score of 0.98. The general results indicate a viable and holistic response that adds to secure interactions, blocks malicious acts and encourages trust in the actual social media settings.
Authors - Md. Shahidul Islam, Hasina Islam Abstract - Cross-domain recommendations are imperative in the growing tourism industry and with the increasing means of communication. Preference drift, preference transfer, and unfamiliarity with places have an overbearing impact on recommender systems. Most approaches do not address geometric misalignment across domains, which is essential for cross-domain preference shift analysis in recommendation tasks. We propose Procrustes-Based Contextual Thompson Sampling (P-CTS) for Cross-Domain POI Recommendation, integrating adversarial domain-invariant learning, optimal geometric alignment via Procrustes transformation, and adaptive Thompson Sampling with sleeping bandit management. First, the embeddings are constructed to model the preference drift across the domains. Next, the Procrustes transformation aligns source and target embedding spaces via optimal rotation, scaling, and translation. In the last phase, we initialize Beta priors with similarity-weighted pseudo-counts derived from the aligned embeddings. The experiments on Gowalla and Foursquare across domains demonstrate 5.1% improvements in Precision@5 and 9.75% improvements in cold-start accuracy, suggesting an adaptive exploration-exploitation trade-off.
Authors - Binh Pham Nguyen Thanh, Long Duong Phi, Phung Thi-Kim Nguyen, Nhan Thi Cao Abstract - The rapid proliferation of Internet of Things (IoT) devices has significantly increased the digital attack surface, which, in turn, has raised network vulnerability to sophisticated Distributed Denial of Service (DDoS) campaigns that could reduce the effectiveness of traditional signature-based Intrusion Detection System (IDS). Furthermore, conventional Machine Learning (ML) approaches are often subject to manual feature engineering and lack the capture of complex spatial and temporal dependencies, which are essential to detect subtle, polymorphic threats. In this regard, the present work proposes a lightweight hybrid Deep Learning (DL) architecture for reliable (DDoS) detection. The proposed approach integrates spatial feature extraction using a Convolutional Neural Network (CNN) with a Bidirectional Long Short-Term Memory (BiLSTM) network to capture temporal correlations, further enhanced by an additive attention mechanism that underlines the importance of flow segments relevant to recognition. To mitigate issues with computational complexity, a two-phase hybrid feature selection approach, a combination of Information Gain (IG) and Dynamic Particle Swarm Optimization (PSO) would be utilized to select an optimal subset of features. The performance of the model was evaluated using the CICDDoS2019 benchmark dataset. The feature selection process was able to reduce the input space from 80 to 17 relevant features. The combined CNN-BiLSTM model, along with threshold optimization, was able to achieve an accuracy of 94.1%, which indicates a significant improvement in the reduction of false negatives and validates the effectiveness of the proposed method in a secure IoT environment.
Authors - Wani Zahidah Mohd Subari, Shuzlina Abdul-Rahman, Mohamad Faizal Ab Jabal, Sharifalillah Nordin Abstract - Role-playing games (RPGs) allow the player to take on a specific role and complete different missions during gameplay. Their diversity enables a range of ap-plications beyond entertainment, as they are often used in educational contexts. Learning content can be embedded in common components, such as game fields, tasks, objects, or non-playing characters (NPCs). The paper presents several educational RPGs with their features and characteristics, and existing models of didactic video games. It proposes a two-level metamodel for describing an educational RPG. The metamodel is divided into five main components (world, educational aspects, quest, playing character, and NPCs), and their taxonomies are presented briefly. The authors propose a conceptual model that includes the interrelationships among the components mentioned. In addition, their interpretations and significance for the development of RPG educational games are explained. An example of the metamodel is represented through a quest from a real educational RPG in the field of Chemistry. The presented RPG metamodel improves under-standing and helps to better design, develop, and integrate such games into various learning environments. The presented taxonomy can serve as a useful template for structuring design details.
Authors - Prabhat Kumar Gupta, Perumal T, Karthick Pannerselvam Abstract - Generation Large language models, as well as retrieval-augmented generation (RAG), are highly performing on semantic queries, but with considerable latency as they require embedding computation, a vector similarity search, and generation at inference time. Such delays make them inappropriate in time-sensitive and domain-specific retrieval activities. In this paper, the Hierarchy Latent Retrieval Model (HLRM) which is a deterministic architecture will be introduced and able to answer semantic queries in O(1) constant time. HLRM unites hierarchical semantic routing and semantic hashing so that pre-validated units of knowledge can be directly illuminated without the need to search methods or language model informing of their existence at run time. All computationally expensive processes are done offline, which means that embedding processes or vector databases are not needed to run a query. Milliseconds-response time with very high exact-match accuracy is proved under experimental assessment on an orderly institutional knowledge environment. The findings suggest that HLRM offers an alternative of fast, interpretable, and reliable systems to the generative retrieval systems in non-random settings where precision and response latency is paramount.
Authors - Khedkar Aboli Audumbar, Uday Pandit Khot, Balaji G. Hogade Abstract - Malicious or compromised internal users can act like normal users with valid login credentials and thus become difficult to detect. As a result of their similarity to normal users, traditional methods of detecting intrusions, have difficulty identifying the subtle and changing behaviors of malicious insiders. This paper introduces a comprehensive User and Entity Behavior Analytics (UEBA) framework to help detect malicious insiders. It works by analyzing activity logs generated by the enterprise. Further it performs data cleaning and feature engineering; creating behavioral profiles for each user based upon the attributes of time, environment, and behavior. These profiles are used to model normal interaction patterns and with the DBLOF algorithm, an outlier score for each profile is created. The outlier score indicates whether or not a given user’s behavior has deviated from normal. In order to make the proposed system adaptable to changing environments over time, it utilizes deep learning algorithms to detect changes in behavior and to increase the accuracy of anomalous behavior detection. The proposed system also enables the ingestion of real-time data, the evaluation of risk, and the display of alerts in a visual format. Thus, providing the scalability and operational performance required to support large-scale organizations. Overall, the proposed system represents a reliable, modular, and understandable UEBA framework. It is capable of accurately detecting malicious insider threats and representing an efficient method for proactively mitigating risks through security operations within enterprises.
Authors - Poonam Chaudhary, Rita Chhikara, Nupur Prakash Abstract - This work addresses the challenge of Isolated Sign Language Recognition (ISLR) on mobile and edge devices, where computational resources, memory, and energy budgets are severely constrained. Existing approaches based on pixel-level three-dimensional convolutional neural networks are computationally expensive and sensitive to background variations, while recurrent models such as Long Short-Term Memory networks suffer from a sequential processing bottleneck that limits parallel execution on modern hardware accelerators. To overcome these limitations, this paper proposes a hybrid Adaptive Graph Convolutional Network (A-GCN) and Transformer architecture that decouples spatial and temporal modeling of skeletal sign representations. The A-GCN employs a learnable adjacency matrix to capture dynamic and semantically meaningful spatial relationships between skeletal landmarks, while the Transformer encoder leverages parallel self-attention to model long-range temporal dependencies without recurrence. Experimental evaluation on the 250-class Google Isolated Sign Language Recognition dataset demonstrates a Top-1 accuracy of 78.90%, outperforming a Bi-LSTM baseline by 6.96%. In addition, the proposed model achieves a throughput of 400.55 frames per second with a latency of 2.50 ms on accelerator hardware and maintained real-time performance on consumer-grade devices. These results demonstrate that landmark-based, parallel architectures enable accurate, real-time, and privacy-preserving sign language recognition suitable for deployment on standard mobile devices.
Authors - Subin Simon, Prathilothamai M Abstract - Deep learning has shown significant potential in medical image classification; however, a systematic comparison of deep feature extraction strategies for multi class diabetic eye disease assessment remains limited. This study presents a comprehensive comparative analysis of seven deep learning architectures, including conventional CNN, pretrained VGG16, Vision Transformer (ViT), Conformer, hybrid CNN ViT, and attention-augmented variants incorporating Squeeze-and-Excitation (SE) and Convolutional Block Attention Module (CBAM). All models are evaluated under a unified preprocessing and training framework to ensure fair performance comparison.The investigation focuses on analyzing how different architectural paradigms capture discriminative local and global retinal features relevant to disease classification. Extensive experiments are conducted on public fundus image datasets using standard evaluation metrics, including accuracy, precision, recall, and F1-score. Experimental results demonstrate that hybrid and attention-integrated architectures outperform standalone CNN and transformer models. In particular, the Conformer architecture achieves the best overall performance, reaching approximately 91% classification accuracy in the four class setting (Diabetic Retinopathy, Glaucoma, Cataract, and Normal), while the CNN ViT model attains approximately 89% accuracy.These findings highlight the effectiveness of combining convolutional operations with global self-attention mechanisms for robust and discriminative feature extraction in automated diabetic eye disease classification.
Authors - M.Murugesen, Priyanka P Abstract - Deep learning–based medical image models have achieved expert level performance in GPU-based research environments [1–3]. However, relia ble deployment in real clinical systems remains challenging due to constraints related to power consumption, hardware stability, and long-term operation. While prior studies have focused on improving model architectures or hardware accelerators [4,5], relatively limited attention has been devoted to systematical ly managing the transition from GPU-based development to NPU-based de ployment environments. This study formulates the GPU-to-NPU transition as an independent deployment research problem. Rather than proposing a new model architecture, we focus on preserving functional equivalence when trans ferring a validated GPU-trained medical vision model to an NPU-based infer ence environment. The proposed framework consists of reference model fixa tion, intermediate representation (IR)-based conversion [13–15], operator com patibility management, inference pipeline alignment, and output-level function al equivalence validation. The framework is evaluated through deployment of a ResNet-50–based pa thology classification model on a commercial ATOM NPU platform. Experi mental results demonstrate a 99.1% agreement rate (991/1,000 samples) be tween GPU-based and NPU-based inference outputs, confirming consistent de cision behavior despite architectural differences. These findings indicate that deployment reliability depends more on execution environment control and preprocessing alignment than on model architecture modification. By redefining deployment as a structured research problem, this work pro vides a reproducible methodology for translating research-grade medical AI models into energy-efficient NPU inference systems under practical clinical constraints.
Authors - Jayanthi J, P.Uma Maheswari, S.Uma Maheswari, Arun Kumar, Karishma V R Abstract - The rapid migration of artificial intelligence from cloud platforms to edge-based Internet of Things environments has intensified the demand for transparent and trustworthy decision-making under severe resource constraints. While edge intelligence enables low-latency and privacy-preserving analytics, the opacity of deployed models limits user trust, accountability, and regulatory acceptance. Existing explainability techniques largely assume cloud-level resources, making them unsuitable for real-time and energy-limited edge deployments. In order to close this gap, this work develops an interpretable intelligence framework that is resource-aware and adaptable, specifically designed for limited IoT systems. The suggested approach integrates interpretability directly into the decision-making process, allowing for the generation of faithful, lightweight explanations in addition to predictions while dynamically adjusting to operational context and runtime restrictions. Further balancing local responsiveness with system- level insight aggregation and secure governance is achieved through hierarchical explanation control. Transparency, efficiency, and scalability are all in line with the framework's treatment of explainability as a fundamental system capacity. The study shows that adaptive, deployment-aware explainability can greatly improve edge intelligence's operational reliability and trustworthiness. These insights establish a foundation for building accountable and interpretable AI systems in real-world IoT environments.
Authors - Rimon Kumer Roy, Jannatul Ferdous, Kazi Lutfur Nahar Mithila, Sabbir Islam, Mohammad Zahid Hassan, Sadah Anjum Shanto Abstract - Early identification of ophthalmic disease is critical to pre serve eyesight. We present DeepEye, a stacking-ensemble framework for multi-disease classification on the Eye Disease Image Dataset (EDID, Mendeley Data). After standardized preprocessing and augmentation, f ive architectures ResNet50, VGG16, DenseNet121, EfficientNet-B4, and Vision Transformer were trained and evaluated. The final ensemble in tegrates the top base models with a logistic regression meta-learner op timized via hyperparameter tuning. On a held-out test set, DeepEye achieves 91.34% accuracy and AUC of 0.9965, outperforming all con stituent models and exhibiting stable gains across cross validation folds. Model transparency is supported with Grad-CAM visualizations that lo calize disease-relevant regions, enhancing clinical interpretability. These results indicate that combining convolutional and transformer backbones within a tuned stacking framework yields a high-accuracy, explainable approach for automated eye disease detection in healthcare settings.
Authors - Nguyen Thi Hoi, Vu Thi Anh Hong, Dang Thi Anh Tho, Dang Thuy Linh, Nguyen Khanh Linh Abstract - The increasing use of renewable energy sources has made the integra tion of Flexible AC transmission system (FACTS) devices into contemporary power systems, an important area of research. The function and effectiveness of FACTS devices in enhancing power quality and preserving stability in traditional power systems and those that significantly count on renewable energy source are comprehensively examined in this study. Variability and unpredictability brought about by renewable energy sources can negatively impact the voltage profile, particularly at high penetration levels. Devices from the Flexible AC Transmis sion System, like the Thyristor-Controlled Series Capacitor (TCSC) & Static Var Compensator (SVC), provide efficient ways to improve system stability and dy namically regulate voltage. This paper investigates a coordinated control strategy of SVC and TCSC for improving voltage profiles in a transmission network with high renewable energy integration. Using an IEEE-14 bus test system, various scenarios of renewable penetration are simulated to analyze the performance of coordinated FACTS operation. The findings show that the suggested coordinated control improves overall system dependability and power transfer capabilities in addition to reducing voltage variations and reactive power imbalances. The study highlights the importance of optimal placement and coordinated tuning of FACTS devices as a cost effective solution for enabling secure and stable opera tion of renewable-rich power grids.
Authors - Josue Piedra, Nelson Piedra Abstract - Accurate crop production forecasting is essential for sustainable agricultural planning, effective resource management, and long-term food security. Conventional statistical and regression-based models often fail to capture the complex, nonlinear relationships that exist among agro-climatic variables, soil characteristics, and crop yield [1]. To address these limitations, this paper proposes an agentic artificial intelligence (AI)–based framework for crop production analysis that integrates autonomous decision-making with machine learning and deep learning techniques. The proposed framework utilizes agro-climatic and soil parameters such as temperature, humidity, soil moisture, cultivated area, and seasonal information to model crop production behaviour. Three predictive approaches— Linear Regression, Random Forest, and CNN–LSTM—are implemented and evaluated within the agentic framework using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and the Coefficient of Determination (R2) as performance metrics. Experimental results demonstrate that the Random Forest model significantly outperforms the other models, achieving an RMSE of 0.56, MAE of 0.31, and R2 value of 0.96. These findings highlight the effectiveness of agent-driven machine learning systems in accurately modelling agricultural data and supporting intelligent decision-making for crop yield optimization.
Authors - Priyanka Halder, Anupam Sinha Abstract - This study analyzes the extent to which credibility from influencers impacts consumers' buying behavior. The focus will be on how the intention to buy impacts this relationship as the problem is being analyzed in the context of social commerce on TikTok. The study is developed within the framework of Source Credibility Theory which suggests that consumers’ perception and consequent behavior are influenced by the perceived degree of the spokesperson’s Attractiveness, Trustworthiness, and Expertise. The study employs a quantitative explanatory methodology. A purposive sampling technique was used to collect data from a sample of 100 active TikTok users who follow the provided influencer. The analyzed relationships will be quantified using Structural Equation Modelling with Partial Least Squares (SEM-PLS). The research results concluded that influencer credibility increases the intention to buy, but does not increase the purchasing decision. The intention to buy completely mediates the relationship between influencer credibility and purchasing decision. This demonstrates that influencer credibility is a significant factor in the intention to buy behavior, but it is the intention that is essential in order to convert the persuasive influence into actual buying behavior. The study contributes to digital marketing communication research by extending Source Credibility Theory to the context of short-video social commerce platforms.
Authors - Swasti Shinde, Ishita Rajarshi, Shravani Mote, Abhilasha Gandhi, Megha Dhotay Abstract - The use of artificial intelligence (AI), especially deep learning, has brought major changes in the study of histopathology images for cancer diagnosis for the doctors. This review compares the latest deep learning methods used to identify whether the tissues are Benign. (noncancerous) or malignant (cancerous). This paper discusses different technologies used for the study of histopathology images. Convolutional neural network (CNNs) is effective in capturing small local details in the images. Several studies report that Transformers (ViTs) can outperform CNNs in complex classification tasks [18, 22]. Along with the discussion about newer hybrid models and large foundation models, the paper specifies the strength of combining the strength of both. Although these models are developed focusing on achieving high accuracy on good data sets, there are few challenges in their practical use, such as i. Models often fail to generalize data from different hospitals due to domain shift [1, 28] ii. Model interpretability remains a significant challenge in clinical development [34]. iii. There is a lack of proper methods to measure the uncertainty in the decisions [40,41]. This paper highlights the research gaps in real clinical use and focuses on the need to develop models that are robust, interpretable, and suitable for practical healthcare applications.
Authors - Ioana Chan Mow, Fiafaitupe Lafaele, Sarai Faleupolu-Tevita, Vensel Chan, Soonalote Eti, Fiti Tolai Abstract - The need to move to online proctored exams urged the National University of Samoa (NUS) to trial and evaluate a variety of online proctoring systems to ensure the offering of safe and secure exams online. The aim of the 4-phase research was to answer the following question: “What are some feasible options for online proctoring systems (OPSs) for offering online exams for NUS?” This paper is based on the last phase of this 4-phase study conducted at NUS to evaluate the feasibility of two proctoring systems, Integrity Advocate and Proctorio, for online exams, particularly during lockdown. Specifically, the objectives were to i) trial and evaluate the suitability of each OPS as well as the type of exam mode (two options: in the laboratory or from home) using a diffusion of innovation framework and, from the evaluation, recommend a suitable OPS for NUS. Both between-subjects and within-subjects analyses revealed highly positive responses for both OPS and exam mode across the 5 variables of the diffusion of innovation model of relative advantage, compatibility, ease of use, observability, and trialability. Most of the findings did not show any differences by OPS type, exam mode, or gender, as most responses across the 5 variables of diffusion of innovation were highly positive and very similar, indicating positive and high rates of adoption of the two OPS. An in-depth investigation into the features of the two OPS also revealed that Proctorio had a wider scope of features than Integrity Advocate.
Authors - Syammas Pinasthika Syarbini, Irmawan Rahyadi, Muhammad Aras, La Mani Abstract - The need to move to online proctored exams urged the National University of Samoa (NUS) to trial and evaluate a variety of online proctoring systems to ensure the offering of safe and secure exams online. The aim of the 4-phase research was to answer the following question: “What are some feasible options for online proctoring systems (OPSs) for offering online exams for NUS?” This paper is based on the last phase of this 4-phase study conducted at NUS to evaluate the feasibility of two proctoring systems, Integrity Advocate and Proctorio, for online exams, particularly during lockdown. Specifically, the objectives were to i) trial and evaluate the suitability of each OPS as well as the type of exam mode (two options: in the laboratory or from home) using a diffusion of innovation framework and, from the evaluation, recommend a suitable OPS for NUS. Both between-subjects and within-subjects analyses revealed highly positive responses for both OPS and exam mode across the 5 variables of the diffusion of innovation model of relative advantage, compatibility, ease of use, observability, and trialability. Most of the findings did not show any differences by OPS type, exam mode, or gender, as most responses across the 5 variables of diffusion of innovation were highly positive and very similar, indicating positive and high rates of adoption of the two OPS. An in-depth investigation into the features of the two OPS also revealed that Proctorio had a wider scope of features than Integrity Advocate.
Authors - Maulikkumar Pandya Abstract - Skin lesion segmentation is essential for computer-aided dermatological diagnosis, but reliable pixel-level annotations are costly and require experts. To reduce dependence on manual labeling, pseudolabeling combined with foundation models such as the Segment Anything Model (SAM) has been explored; however, most pipelines rely on a single pseudo-label per image, which can introduce boundary bias when pseudo-labels are noisy. In this paper, we compare two U-Net training pipelines built on pseudo-labels generated using U²-Net and SAM. The first pipeline follows a single pseudo-label inheritance strategy as a strong annotation-free baseline. The second pipeline synthesizes multi-style pseudo-labels (tight/moderate/loose) and applies agreement-based learning to supervise only high-confidence consensus regions while suppressing uncertain boundary pixels. No ground-truth masks are used during training; manual annotations, when available, are used only for offline evaluation. Experiments on ISIC 2018 under a pseudo-reference protocol show improved boundary behavior (higher Boundary F-score) and more coherent contours, especially in ambiguous border regions.
Authors - Satyendra Sharma, Pradeep Laxkar Abstract - Reconstructing polyphonic musical sequences represents a significant challenge in computational music analysis. This study presents a method based on empirical entropy and the analysis of multi-voice bigrams to identify and re-construct missing notes in polyphonic sequences. The approach combines statistical modeling of transitions between simultaneous voices in a musical piece, represented as tuples duration:interval|duration:interval|... depending on the number of voices, with techniques for generating and ranking possible segments according to probability and entropy. Results show that considering multi-voice bigrams effectively captures the polyphonic structure and improves the accuracy of missing note prediction. This work opens new perspectives for the application of probabilistic models to polyphonic music and AI-assisted music generation.
Authors - Paponsun Eakkapun, Sulak Sumitsawan, Chukiat Chaiboonsri Abstract - Cloud cover, shadows, haze, illumination variation, and atmospheric noise severely degrade the reliability of satellite image change detection. This paper proposes an atmosphere-aware, physics-driven preprocessing framework that performs cloud, shadow, haze, and illumination compensation before change analysis, without relying on convolutional or transformer-based networks. Two multi-temporal satellite images are processed through unified cloud and shadow handling, haze correction, illumination normalization, and residual atmospheric noise suppression, followed by a spectrally invariant change detector with structural consistency validation. The system also generates semantic multi-class change maps and geo-contextual text explanations to enhance interpretability. Experiments on diverse multi-temporal datasets demonstrate a change detection accuracy of 98.9% with high precision and recall, significantly outperforming conventional and deep learning baselines
Authors - C. R. Patil, Arundhati Sarvadnya, Diksha Shejwal, Sakshi Nehe, Sobiya Shaikh Abstract - The rapid expansion of the Internet, together with the pervasive diffusion of mobile technologies, has fundamentally reshaped contemporary socio-economic activities, positioning e-commerce as a core pillar of the digital economy. In response to increasing competitive pressures and the growing demand for personalized consumer experiences, enterprises have progressively adopted advanced analytical technologies, among which machine learning has emerged as a key strategic instrument. This study develops and empirically evaluates a machine learning–based product recommendation framework that integrates historical transaction data with sentiment information extracted from user-generated reviews. Data were collected from multiple e-commerce platforms and assessed using widely adopted evaluation metrics, including Accuracy, Recall, and F1-score. The experimental findings demonstrate that the XGBoost algorithm consistently outperforms alternative models, exhibiting superior capability in identifying latent consumer preferences and behavioral patterns. Overall, the results provide robust empirical evidence supporting the effectiveness of the proposed approach and underscore its practical potential for enhancing personalization quality and improving recommendation performance in large-scale e-commerce environments.
Authors - Upendra Pratap Singh, Akshay Anand Abstract - The rapid proliferation of Internet of Things (IoT) systems has led to the widespread adoption of artificial intelligence for autonomous sensing, prediction, and decision-making across critical application domains. While these AIdriven IoT systems achieve high operational efficiency, their increasing reliance on complex and opaque models raises serious concerns regarding transparency, trust, accountability, and regulatory compliance. These concerns are particularly acute in distributed IoT environments, where decisions are made across heterogeneous devices under resource constraints. Existing explainable artificial intelligence (XAI) approaches largely focus on centralized or standalone machine learning models and fail to address the unique challenges of IoT systems, including deployment heterogeneity, dynamic data distributions, privacy requirements, and real-time decision-making. As a result, explanations are often disconnected from system behavior, lack consistency across layers, and provide limited support for trust assessment and human oversight. This paper presents a comprehensive survey of explainable AI techniques for trustworthy IoT systems and introduces a deployment-aware reference architecture that integrates explainability, trust evaluation, privacy preservation, and human-in-the-loop feedback across edge, fog, and cloud intelligence layers. The architecture emphasizes localized explanation generation, context-aware refinement, explanation validation, and multi-metric trust assessment, enabling explanations to evolve alongside system behavior. By explicitly coupling explanation quality with trust monitoring and adaptive feedback, the proposed framework bridges the gap between predictive performance and operational trustworthiness in distributed IoT environments. The survey highlights key research trends, identifies critical gaps in current methodologies, and outlines future directions for scalable, reliable, and human-centered explainable IoT systems. By positioning explainability as a core system property rather than a post-hoc add-on, this work provides a foundation for designing AI-enabled IoT systems that are transparent, accountable, and trustworthy by design.
Authors - Sreebala V S, Arun Kumar V N, Agna.S. Nath Abstract - The Commercial Territory Design Problem (CTDP) plays an important role in sales and marketing management. The problem focuses on partitioning some basic units into territories to optimize compactness while ensuring workload balance and connectivity constraints. Due to the NP-hard property of the problem, exact approaches often have limitations in scalability across large datasets. This study proposes a combination of the classical ALNS algorithm framework and an ActorCritic Deep Reinforcement Learning architecture to deal with the large CTDP instances. Our proposed algorithm can automatically select destroy and repair operators, and dynamically fine-tune hyperparameters such as destruction level and acceptance criteria based on the actual state of the search process. Experimental results on benchmark instances with various sizes show that our algorithm not only achieves superior quality solutions compared to traditional ALNS but also surpasses exact solutions in terms of convergence speed within the same runtime limit. It can achieve high-quality solutions within a reasonable execution time and has the potential for real-world applications.
Authors - Tri Wiyana, Roberto Tomahuw Abstract - Mental health disorders are among the major global health problems, and early diagnosis is the key for effective management. Conventional methods are based on self-reported or clinical scales, for which intervention comes late. In this paper, we propose a multimodal AI framework for the detection of early mental health detection from typing and voice behaviors. We extract BERT-based linguistic embeddings of text transcripts and spectral features of the speech signals from the audio data using the DAIC-WOZ dataset for capturing verbal cues. These features are then combined by machine learning algorithms to classify depression. The proposed framework prioritizes non invasive, privacyconscious detection with explainability techniques used to foster clinical confidence. We further present experimental results to show that the multimodal fusion also provides classification gain over unimodal baselines. This study demonstrates the capability of AI-based, real-time methods for proactive mental health monitoring and provides a stepping stone towards healthcare deployment.
Authors - Shylaja P, Jayasudha J S Abstract - The Question Answering system (QA) is one of the popular and widely used ap-plications of NLP. It is an information retrieval system that attempts to find the correct answer for a question based on the given paragraph text. Transformers have been widely used for QA tasks, due to their contextual embedding, attention mechanism, and transfer learning for specialized tasks. Transformer-based models can be easily fine-tuned with large datasets. Such models provide state-of-the-art performance over large datasets for question-answering tasks. The proposed approach compares performance of transformer based model over a small sized dataset. We incorporated an answer formation layer along with transformers to comprehend contextual, syntactical, and semantic information from small-sized datasets. We developed a set of rules according to question categories to generate semantically and syntactically coherent option sentences based on the questions. These rules transformed option phrases into contextually appropriate sentences. We evaluated SBERT transformer models namely all-mpnet-base-v2, all-MiniLM-L6-v2, all-distilroberta-v1 over answer formatted data and it showed in-crease in accuracy. Answer formation rules over noun phrases of small-sized datasets can help transformers to learn contextual knowledge about the options in the QA sample, Addition of answer formation stage on samples of SciQ data resulted in a rise in accuracy from 86 % to 92 % when using all-MiniLM-L6-v2 model.
Authors - Asritha Paruchuri, Gudivada Krishna Prakash, Mulla Junaid Rahman, Lambu Damarukanath, Guttikonda Prashanti Abstract - The sharing of images in decentralized settings needs high assurances of secrecy, integrity and controlled access. The fast development of cloud-based services and online communication tools have multiplied the communication of sensitive images, and the traditional centralized storage and single-layer security systems are susceptible to cyber-attacks, unauthorized access, and data leakage. The presented paper outlines a safe and decentralized image-sharing system based on Advanced Encryption Standard (AES), the Secret Sharing scheme by Shamir, blockchain authentication, and decentralized storage with the Interplanetary File System (IPFS). First, the input image is encrypted with the help of AES to provide high cryptographic confidentiality. The ciphertext image is further split into shares with secret sharing scheme that avoids unauthorized disclosure and only allows the reconstruction of the encrypted image when the necessary number of valid shares is received. The encrypted shares that are generated are stored in a decentralized way using IPFS, which is highly available, fault tolerant, and does not have a single point of failure. Decentralized access control, participant authentication and integrity verification that is tamper-resistant are enforced using blockchain technology. In the reconstruction process, the encrypted image is reconstructed with the help of Lagrange interpolation and then decrypted with the help of AES, which guarantees safe and lossless recovery of the original im-age. The suggested framework offers multi-layer security, increases confidentiality and data integrity, removes centralized vulnerabilities, and is highly resistant to unauthorized access and data-alteration.
Authors - Mach Thai Loc, Nguyen Hong Phuc, Huu-Cuong Nguyen Abstract - With the development of e-commerce and global supply chains, there is a growing concern about fake or counterfeit products. Current methods for verifying product authenticity are often cumbersome ,time consuming, and vulnerable to tampering. In order to address these issues, for the purpose of this project, a QR code based "Fake Product Detection System" is introduced. In this system, the manufacturer creates an exclusive QR code for each product. The manufacturer then keeps the QR code in a database. If the QR code is scanned through the web-based application, the code is instantly verified. If the code is unique and has not been used be-fore, the product is genuine. But if the code is duplicated or used multiple times, the product is deemed counterfeit. The system is implemented using the Flask web development framework, SQLite database, and web interfaces using the HTML/CSS duo, which is lightweight and easy to use. Other notable features of the system are user authentication, history logging, suspicious image upload for the QR code, and detection of counterfeit items. Overall, this solution would offer a simple, economical, and efficient means to uncover Trojan products while fostering trust amongst consumers and aiding manufacturers to track counterfeit practices.
Authors - K. Thirupathi Reddy, K. Venkata Ajay Kumar, M. Kaveri Abstract - This study explores female creators’ subjective lived experiences navigating human–AI interaction (HAI) within generative design ecosystems. It examines how creators engage with intelligent systems during collaborative creation processes and how they negotiate creative agency between algorithmic outputs and personal meaning-making. Drawing on an Interpretative Phenomenological Analysis (IPA) approach, the study involves seven women who actively utilize Canva’s AI-enabled capabilities to produce professional digital content. Data were collected through in-depth semi-structured interviews and digital observation of design outputs distributed on Instagram. The findings indicate that participants interpret Canva AI as a collaborative creative partner that supports iterative dialogue, experimentation, and reflective decision-making. Rather than replacing human authorship, AI interaction functions as a mediated process in which creators provide prompts, reinterpret generated results, and refine instructions to align outcomes with their subjective intentions. This interaction fosters a sense of psychological safety, particularly among non-professional designers, enabling them to explore creative practices with greater confidence. Through this ongoing negotiation between human agency and algorithmic assistance, participants describe pathways toward professional identity formation and increased participation in contemporary digital creative cultures. Overall, the study highlights how intelligent design systems can shape meaning-making processes, reinforce creative self-efficacy, and support women’s evolving roles within AI-assisted visual communication practices.
Authors - Dudi Gnana Prasoona, Zeenathunnisa, Yamuna V, Pushyami B, Ramandeep Kaur, Navjot Kaur Abstract - In the global health sector, Diabetes is a major concern which needs accurate and effective models for early prediction. This work is quantitative re-search work. The dataset was collected from CDC Diabetes Health Indicators, and we used Light Gradient Boosting Machine (LightGBM) model for predicting diabetes. Since this research work is binary classification-based work, in our data preprocessing stage, we used Synthetic Minority Oversampling Technique (SMOTE) for controlling class imbalance and for feature selection we used Chi-square test to improve the model performance. The proposed LightGBM model showed its ability to recognize complex correlation between diabetes-related health indicators with the training accuracy of 92% and a ROC-AUC score of 0.97 on the test dataset. Overall, the findings highlight that predictive accuracy is significantly improved after applying both imbalance data controlling and most correlated feature selection techniques.
Authors - Lekshmipriya Vijayan, Bindu V R Abstract - In the present paper, a model on an EOQ policy for deteriorated inventory items with stock-sensitive demand pattern under inflation when the deterioration rate is considered to be a linear function of time. Partially backlogged shortages form is allowed to occur in this system. The required conditions are stated to validate the optimal solutions of the present model. Furthermore, the average cost function and decision variables such as shortages time-point and replenishment cycle have been computed with the help of a step-by-step solution procedure and Mathematica software 12.3. Finally, a numerical example as well as its post-optimal analysis for theoretical model is presented to illustrate the pro-posed work.
Authors - Windy Permata Suyono, Marsellisa Nindito, Dwi Handarini, Ratna Anggraini, Eka Septariana Puspa, Surya Anugrah, Sabo Hermawan, Rio Firnanda, Irima Rahmadani Abstract - Gas leaks pose a threat to safety because they can cause fires and damage to property, and they are sometimes fatal. Traditional detection methods are manually dependent or delayed in response, which means they are not always reliable and timely. This paper presents Safe- Gas system for gas leak detection, monitoring and automatic shut-off. The system uses low-cost gas sensor, flame sensors, load cell and an ESP32 microcontroller for local processing. The system is connected to the cloud via Firebase to send alerts, and it has a battery backup to keep it running when the internet or power goes off. The app supports both remote and autonomous valve shutoff. SafeGas is a name that stands for resilience and accuracy. The designers and developers of the device have tested it in the laboratory and in the field to ensure it meets the set standards. First, the system aims to reduce the number of false alarms. Second, it can operate without an internet connection. Third, it can take safety measures independently. The embedded system and cloud integration aspects of the project demonstrate how they can be combined to achieve the desired results.
Authors - Sachin Ratnaparkhi, Parikshit Mahalle, Pankaj Chandre Abstract - Spatial judgment, incorrect furniture size, and poor personalized decor advice are common issues in most interior design planning. The aim of this paper is to introduce an AI-powered Augmented Reality Interior Design Assistant that makes it possible for users to visualize furniture and decor in real spaces using accurate real-world measurements. Spatial mapping using SLAM based AR core plane detection and depth sensing allows for more accurate estimations in room sizes, identifies objects in the scene, and makes AI-driven suggestions on furniture size and styles. .A hybrid AI engine is built using K-nearest neighbours, collaborative filtering and feature extraction methods. The AR rendering process takes care of depth by modifying 3D assets to expected sizes to make sure everything is placed correctly. The AR application is based on Unity 3D with AR Foundation and ARCore, the backend services are provided by python(flask) connected through RESTful APIs, for user profile and catalog management Firebase/PostgreSQL is used. Scikit is used for building machine learning models which is supported with Numpy and Panda for data handling. The assistant will also provide design tips through a conversational AI feature that makes it accessible to everyone. Tests show a significant reduction in spatial errors, much faster design decisions, and better relevance of recommendations. These results indicate that real-scale visualization with AI suggestions tremendously increases design confidence and at the same time reduces the need for redesigns. This system connects AR visualization with AI interior support for a smooth and smart design experience.
Authors - Deepikaa R Ra, Sriram S, Sudhanthira G Abstract - Huntington’s disease is a devastating brain disorder. It gradually destroys nerve cells due to mutations in the HTT gene that disrupt gene functions. Years of research have not led to effective treatments that can slow or stop the disease. Clearly, we need faster ways to find new drugs. This paper introduced an AI-powered systems biology framework that examines both transcriptomic and clinical data to identify drugs that could be repurposed for Huntington’s disease. First, it uses ordinary least squares regression to remove any unusual variables followed by creating gene co-expression networks to closely examine the specific molecular disorder in the disease. Next, they conduct differential network analysis to identify pathways and transcriptional regulators that go awry and compare known drug effects with Huntington’s molecular signatures, rating each drug based on its ability to reverse those harmful gene changes. This helps them quickly focus on drugs that might actually be effective. The entire setup allows researchers to filter, rank, and test potential treatments efficiently, improving the process's reproducibility and reliance on real data.
Authors - Vani K S, Nanditha B, R Bharadhwaj, Rishika Ghai Abstract - Internet of Things (IoT) applications have experienced fast development resulting in massive interconnectivity of devices, and IoT networks have become susceptible to security risks of Sybil, flooding, and masquerading attacks. Conventional centralized security schemes lack flagella, lack the dynamism of trust evaluation, and are vulnerable to single-point failures, whereas the current blockchain-based systems impose too much extra computational and energy load to be applicable in resource-constrained IoT applications. These issues underscore the necessity to have a lightweight, decentralized, and trust-conscious security system that can be used to guarantee secure IoT communication in adversarial environments. The paper presents a lightweight framework of blockchain-based trust that can be exploited to provide security to IoT communication against network-level attacks. The suggested architecture combines a decentralized blockchain architecture and dynamic trust assessment operation to distinguish trustful nodes and isolate bad actors. It uses a trust-sensitive Proof-of-Work (PoW) architecture to verify block authenticity, in which a node trust score is calculated following communication behavior and history of interaction. Technique of order of preference similarity to Ideal solution (TOPSIS) is used to choose the high trust nodes to validate the transaction securely, which minimizes the amount of computation wasted and increases the network reliability.
Authors - Palungbam Roji Chanu, Venkata Sathish Kumar Badithala, Nepholar Chongtham, Arambam Neelima, Gulshan Gupta, Rohita Tyagi Abstract - Quantum computers are a major threat to the existing encryption mechanisms. In terms of security, the traditional encryption algorithm depends on complex problems like discrete logarithm as well as factorization of integer. Shor’s algorithm is believed to break the current Public Key Encryption algorithms such as Advanced Encryption Standard (AES). Therefore, several research are carried out in the area of PQC (Post Quantum Cryptography). PQC are based on very complex mathematical problems like Learning with error (LWE) which are robust against quantum computers. The National Institute of Standard and Technology (NIST) has initiated several rounds of standardization process for PQC algorithms, among which NTRU, SABER, CRYSTAL-KYBER are the leading candidates. CRYSTALS-KYBER (Kyber) is the first chosen PQC for standardization. This works explores the recent development in Crystals Kyber implementation and its optimization. Researchers can approach for new research challenges and target for improvement thereby increasing efficiency.
Authors - Taslima Ferdous Supty, Fahima Hossain, Era Aich, Ananna Datta, MD Sahadat Hossen Tanim Abstract - The Newborns mostly use infant crying as their main form of communication and it represents a great variety of physiological and emotional conditions. Despite the high potential of automated infant cry analysis in early diagnosis and support of caregivers, the application in real-life still has low usage rates because of environmental noise, imbalance of classes, low interpretability, and high computational cost. This paper is a compilation of an effective, interpretable, and real-time infant cry classification system using a two-step hierarchical methodology. The first stage involves a distinction of cry and non-cry sounds to reduce the rate of false alarms due to background noise. The second stage involves categorizing detected cries into a particular intent. An adaptive feature fusion strategy based on reinforcement learning, gives the cepstral and prosodic and qualitative acoustic features dynamic weights, and SHAP-based explainability offers explicit feature interpretations. Data augmentation, SMOTE-Audio, and model pruning are used to find solutions to the issues of class imbalance, noise robustness, and deployment constraints. Experimental evidence shows that the proposed approach outperforms single feature base-lines, it is also stable in noisy environments and also attains significant parameter reduction without significant loss in performance, making it possible to run in resource-constrained devices in real time. The system is tested on a publicly available infant cry dataset which contains 889 audio samples of cry and non-cry signals in five categories of cry intent and was recorded in realistic conditions.
Authors - S.Nagarjuna Reddy, B.Lakshmi Priyanka, E.Vamsi, G.Raja Shekar Reddy Abstract - Cloud cover, shadows, haze, illumination variation, and atmospheric noise severely degrade the reliability of satellite image change detection. This paper proposes an atmosphere-aware, physics-driven preprocessing framework that performs cloud, shadow, haze, and illumination compensation before change analysis, without relying on convolutional or transformer-based networks. Two multi-temporal satellite images are processed through unified cloud and shadow handling, haze correction, illumination normalization, and residual atmospheric noise suppression, followed by a spectrally invariant change detector with structural consistency validation. The system also generates semantic multi-class change maps and geo-contextual text explanations to enhance interpretability. Experiments on diverse multi-temporal datasets demonstrate a change detection accuracy of 98.9% with high precision and recall, significantly outperforming conventional and deep learning baselines
Authors - Shweta B. Barshe, Garima B. Shukla Abstract - The use of artificial intelligence (AI), especially deep learning, has brought major changes in the study of histopathology images for cancer diagnosis for the doctors. This review compares the latest deep learning methods used to identify whether the tissues are Benign. (noncancerous) or malignant (cancerous). This paper discusses different technologies used for the study of histopathology images. Convolutional neural network (CNNs) is effective in capturing small local details in the images. Several studies report that Transformers (ViTs) can outperform CNNs in complex classification tasks [18, 22]. Along with the discussion about newer hybrid models and large foundation models, the paper specifies the strength of combining the strength of both. Although these models are developed focusing on achieving high accuracy on good data sets, there are few challenges in their practical use, such as i. Models often fail to generalize data from different hospitals due to domain shift [1, 28] ii. Model interpretability remains a significant challenge in clinical development [34]. iii. There is a lack of proper methods to measure the uncertainty in the decisions [40,41]. This paper highlights the research gaps in real clinical use and focuses on the need to develop models that are robust, interpretable, and suitable for practical healthcare applications.
Authors - Ch.Venkata Narayana, T.Jhansi, D.Charan, K.Priskilla, D.Tejaaswani Abstract - This work proposes an intelligent system for automatic food-image-based recognition and calorie estimation to meet the emerging demand for accurate dietary monitoring and personalized nutrition recommendations. Conventional food-logging methods are cumbersome, prone to errors, and mostly fail to capture portion sizes, hence motivating an end-to-end computer vision and depth-based approach. The proposed system utilizes a custom-curated Indian food image dataset of eighty classes, collected, labeled, and preprocessed to make it robust enough to present various variations in lighting, background, etc. A deep learning model was then trained for detecting and classifying food with high precision. The overall classification accuracy achieved by the proposed system is ninety-seven percent. The depth understanding of the detected food regions will provide an approximation of volume and weight, leading to relatively better calorie calculations. Nutritional analysis gets integrated into the system by relating the type and estimated weight of food to the standard nutritional information for detailed insights in terms of calories, proteins, fats, car-bohydrates, fiber, and micronutrient content. The results for evaluation reveal strong detection, minimum estimation error, and efficient real-time processing, which clearly show its applications. In this paper, an approach that combines recognition by image, depth estimation by portion, and nutrition logic capable of leading to a strong solution for diet determination has been introduced.
Authors - Dipti Varpe, Gouri Kulkarni, Anish Sontakke, Anuj Patil, Prasanna Kekare Abstract - Inconsistent medication intake is a major issue, especially for elderly individuals and patients with memory problems [1]. The MediMitra: Voice Enabled Medicine Alert System seeks to tackle this problem by offering an automated, low-cost and user-friendly medication reminder solution. The system combines Raspberry Pi with Optical Character Recognition (OCR) technology to pull medicine names, dosage details and intake times directly from scanned prescriptions. This reduces manual input and user reliance. The information is stored in a central database and connected to a scheduler that sends timely voice alerts through smart speakers or Bluetooth devices. This ensures users receive reliable and easy-to-access reminders. The OCR module is designed for high accuracy in processing printed prescription images by using image preprocessing techniques like noise reduction and thresholding, which helps in effectively extracting key medication details [2]. The system focuses on accessibility, affordability and ease of use in home or clinical settings. Overall, MediMitra provides a useful technological solution to improve medication adherence and supports independent living. It also has potential for future integration with health-monitoring systems.
Authors - Mamy Haja Rakotobe, Remy Courdier Abstract - This article presents a neuro-symbolic modelling approach grounded in qualitative data collected from 25 sports clubs located in R´eunion. The study develops a methodological chain linking structured semantic extraction, ontological formalisation in OWL, and agent-based simulation implemented in NetLogo. Rather than modifying structural scenarios across experiments, the design introduces two contrasting organisational sensitivity profiles derived from field observations: a damped profile and a high-gain profile. The structural configurations remain identical between profiles; only the coefficients of the commitment update function vary, ensuring strict experimental comparability. Results indicate that identical structural conditions produce differentiated collective trajectories depending on internal sensitivity parameters. In highgain configurations, dominance-weighted interactions increase variance and generate polarised engagement distributions, whereas damped configurations maintain relative stability across scenarios. These findings suggest that modelling organisational sensitivity parameters is critical for understanding the robustness of digitally mediated collaboration in volunteer-based organisations.
Authors - Silvio Simani Abstract - This paper presents a hybrid diagnostic approach for an engine air-path benchmark characterised by environmental variability, limited labelled faults, and the need for reliable online decisions. The proposed method combines physics-guided residual features with datadriven temporal representation learning. Residuals derived from grey-box relations capture physically meaningful deviations, while a lightweight encoder extracts temporal patterns across operating regimes. To enhance robustness under changing ambient conditions, the model is explicitly conditioned on measured environmental variables and trained to favour stable representations across sessions. An open-set decision policy with calibrated rejection is incorporated to reduce misclassification when encountering unseen fault magnitudes or insufficient evidence. The method is evaluated under the official benchmark protocol using online processing constraints and standard metrics, including false alarm rate, detection rate, isolation rate, detection delay, and computational cost. Results show improved reliability compared to competitive baselines, with lower false alarms, higher detection and isolation performance, and stable behaviour across sessions. The approach remains computationally efficient and suitable for real-time deployment in practical diagnostic pipelines.
Authors - Zubayer Bin Ahamed, Umair Hossain, Umara Binte Masud, Abdullah Al Mamun, Md. Rohan Islam, Sadah Anjum Shanto Abstract - Gas leaks pose a threat to safety because they can cause fires and damage to property, and they are sometimes fatal. Traditional detection methods are manually dependent or delayed in response, which means they are not always reliable and timely. This paper presents Safe- Gas system for gas leak detection, monitoring and automatic shut-off. The system uses low-cost gas sensor, flame sensors, load cell and an ESP32 microcontroller for local processing. The system is connected to the cloud via Firebase to send alerts, and it has a battery backup to keep it running when the internet or power goes off. The app supports both remote and autonomous valve shutoff. SafeGas is a name that stands for resilience and accuracy. The designers and developers of the device have tested it in the laboratory and in the field to ensure it meets the set standards. First, the system aims to reduce the number of false alarms. Second, it can operate without an internet connection. Third, it can take safety measures independently. The embedded system and cloud integration aspects of the project demonstrate how they can be combined to achieve the desired results.
Authors - Shafa Salsabila Risfi Febrian, Ricardo Indra, Aura Meivia Safira Arsya, Aurellia Arthamevia Aisyah Abstract - This study examines the determinants of continuance intention in YouTube live streaming consumption among Indonesian Generation Z, focusing on social interaction, entertainment, passing time, and enjoyment. Drawing upon Uses and Gratifications Theory and Computer-Mediated Communication, this research situates live streaming as an interactive digital environment where audiences actively negotiate social and emotional experiences. A quantitative explanatory survey was conducted among 108 Generation Z subscribers of the Windah Basudara YouTube channel, and the data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). The findings reveal that social interaction and passing time significantly influence continuance intention, whereas entertainment and enjoyment do not demonstrate significant effects. These results suggest that sustained engagement in live streaming environments is driven more by interactive and habitual gratifications than by purely hedonic motivations. By highlighting the contextual dynamics of Indonesian gaming live streaming, this study extends the application of Uses and Gratifications Theory in synchronous digital media settings and offers practical implications for content creators seeking to strengthen audience retention strategies.
Authors - Rashmi Vipat, Priyank Doshi Abstract - Agriculture plays a vital role in ensuring food security, yet traditional crop selection and yield estimation practices often fail to account for complex interactions among soil, climatic, and environmental factors. Recent advances in machine learning (ML) have shown significant potential in addressing these challenges by enabling data-driven decision support for farmers. This paper presents a comprehensive review of machine learning–based crop recommendation and yield prediction techniques, focusing on their effectiveness in improving agricultural productivity and sustainability. The study analyzes various supervised and ensemble learning models applied to soil quality parameters such as nitrogen, phosphorus, potassium, pH, moisture, and climatic variables. Emphasis is placed on multimodal data integration, highlighting how the fusion of soil, weather, and remote sensing data enhances prediction accuracy. The review also discusses current limitations, including data scarcity, model generalization, and real-time deployment challenges, particularly in resource-con-strained farming environments. Finally, the paper identifies key research gaps and future directions toward developing robust, scalable, and intelligent agricultural decision-support systems.
Authors - Amit Kalita, Himashree Kalita, Manjit Kalita, Abhijit Chakraborty, Kalpita Dey, Prajukta Deb Abstract - The significance of M- Health platforms to promote health equity has reached critical levels as digitalization in the healthcare sector continues to grow post pandemic. M-Health platform utilization in developing countries like Bangladesh has unique challenges: inconsistent adoption of the digital healthcare system, thus leading to a suboptimal delivery of healthcare services to customers. Using blended models i.e., Expectation-Confirmation Model (ECM), UTAUT2, and the DeLone & McLean IS Success Model, with Training on Virtual Consultation Skills as the moderating variable, the study intends to examine the adoption intention of healthcare providers to continuously use M-Health Platforms for a myriad of services like virtual consultation, remote patient monitoring, electronic prescriptions, and e-health record keeping. This study used Partial Least Squares Structural Equation Modeling (PLS-SEM) to evaluate 898 responses. Social influence, relative advantage, regulatory clarity, digital literacy, trust in technology, and system quality, which collectively improve doctors’ satisfaction with virtual consultation platforms, were identified as important to the purpose of the study. The results offer concrete steps that healthcare providers, platform creators, and policymakers can take to build and improve a solid and dependable M-Health platform that encourages sustained partnership with physicians by alleviating resistance that physicians may have about M-Health platforms in comparable developing countries.
Authors - Yaram Srinivasa Reddy, Bairoju Sreelatha, Shankar Lingam. M Abstract - Knowledge from a resource-rich source domain is leveraged in traditional transfer learning to enhance classification in a relatively data-scarce target domain. However, the resulting target models often suffer from overfitting and limited generalization, which restricts their utility in noisy and resource-constrained environments such as remote sensing. To mitigate these limitations, this work introduces a nuclear norm–regularized teacher–student framework for hyperspectral scene classification. In particular, the student model is regularized with the nuclear norm to encourage low-rank parameter representations, improving robustness to ambient noise. Further, we introduce a relative reconstruction loss (RRL) metric to measure the robustness of the student model to environment noise. Trained on several benchmark datasets, the proposed student model attains up to 87.0% classification accuracy on the independent test sets of UC Merced and EuroSAT, while remaining substantially lighter than the teacher network. Further, relative reconstruction values are computed for different amounts of noise added in the input space; RRL saturate to values less than 1.0 for all the datasets, substantiating that the regularized student model is indeed robust. The competitive performance of the regularized student model compared to the teacher network, its lightweight design, together with RRL values less than one, suggest that the proposed student model can effectively be deployed in noisy and resource-constrained environments such as edge and fog devices.
Authors - Nagesh Sharma, Priyanka Yadav, Kavita Singh Abstract - An accurate determination of childhood malnutrition is necessary for preventive measures. This paper proposes a modified scoring scheme comprising two new elements: the Integrated Anthropometric Score (IAS) and the Hybrid Integrated Score (HIS). IAS uses six primary anthropometric measurements, such as BMI, MUAC, WHZ, WAZ, HAZ, and skinfold thickness, along with selected interaction terms that capture the non-linear connections between growth parameters. The weights are determined by regularized logistic regression, allowing the score to be transparent while still adapting to the statistical structure of the data set. To further stabilize the predictions, the HIS combines BAI, IAS, and a machine learning probability component to make the predictions robust in both synthetic and real-world samples. The models were developed using a synthetic dataset of 9,456 children and tested with five-fold cross-validation and a separate real-world dataset of 38 children. Interaction selection and regularization were performed to control noise sensitivity and avoid overfitting. The findings indicate that the IAS model outperforms BAI with its higher cross-validated accuracy (0.93) and strong performance on real data (0.95). The HIS stays consistent in accuracy across areas and indicates better generalization. The results suggest that by combining multidimensional anthropometric characteristics, interaction-aware modeling, and hybrid learning, a new, more adaptable, and clinically interpretable tool for predicting nutritional risk has been developed, surpassing traditional composite indices.
Authors - Mehak Mukesh Agrawal, Saumya Kumari, Gaganam H V S M Soma Sai, Ankit A. Bhurane Abstract - Most existing artificial intelligence (AI) based assistants are cloud-dependent and require constant internet connectivity. User data is sent to external servers for processing. While this data is often encrypted, it is prone to risks such as cloud security threats. Additionally, users need to be cautious not to share sensitive information. To overcome the aforementioned privacy and internet availability concerns, this paper proposes a completely offline, on-device, cross-device, and open-source system to ensure complete data privacy. The proposed system was tested with several datasets, including AI2 Reasoning Challenge, SQuAD 1.1, CoNLL 2003, GSM8K and StrategyQA to evaluate the closed-form question answering (QA), contextual understanding, named entity recognition, mathematical reasoning and truthfulness, respectively, and with five on-device large language models (LLMs), including Gemma3 1B, SmolLM 1.7B, Qwen2 1.5B, TinyLlama-1.1B, and Phi-2. The system achieved the highest score for closed-form accuracy of 1.0. Its performance on reasoning ranged from 0.01 to 0.23. Truthfulness scores ranged from 0.24 to 0.59. High F1 scores for named entity recognition ranged from 0.74 to 0.79, and contextual understanding scores ranged from 0.02 to 0.17 across the different LLMs. The average response time of the system on mobile and desktop devices was evaluated and observed to vary according to system capability and model size. The system allows users to choose between multiple wake words specific to the Indian context. The proposed system functions on limited RAM and in constrained resource environments.
Authors - Mahzuzah Afrin, Rajasree Das Chaiti, Gazi Tahsina Sharmin Jahin, M. M. Musharaf Hussain, Mohammad Shamsul Arefin Abstract - Reliable identification of pneumonia from chest radiographs plays a central role in supporting clinical decision-making and patient management. Although deep learning models have shown favourable results for automated diagnosis, most existing studies rely on fully supervised training and mainly evaluate performance using accuracy or ROC-AUC metrics. Such evaluations may fail to capture clinical decision reliability, particularly in imbalanced medical datasets. In this work, we examine the effectiveness of self-supervised learning (SSL) for chest X-ray pneumonia classification through a controlled empirical study. A contrastive pretraining strategy is used to learn image representations from unlabeled chest X-rays, followed by supervised linear evaluation. The SSL-pretrained model is compared with a fully supervised model trained from scratch under identical experimental conditions. Our experiments reveal that the supervised baseline attains a slightly higher ROC-AUC; however, this improvement comes at the cost of increased false positive predictions, leading to lower overall accuracy. In contrast, the SSL-pretrained model exhibits a distinct prediction pattern. It achieves higher accuracy and notably improved precision and F1-score, indicating more balanced and reliable predictions. Precision– recall analysis further demonstrates the advantage of SSL in reducing false positive decisions. In addition, Grad-CAM visualizations suggest that the SSL-pretrained model focuses on clinically relevant lung regions. From a clinical decision-making perspective, these results suggest that self-supervised learning offers tangible advantages for chest X-ray analysis when prediction reliability is prioritized. This distinction is especially relevant in clinical settings.
Authors - Prerna Agarwal, Pranav Shrivastava, Samya Ali, Sachit Dadwal, Shubh Om Yadav, Saquib Hussain, Kareena Tuli Abstract - Most existing artificial intelligence (AI) based assistants are cloud-dependent and require constant internet connectivity. User data is sent to external servers for processing. While this data is often encrypted, it is prone to risks such as cloud security threats. Additionally, users need to be cautious not to share sensitive information. To overcome the aforementioned privacy and internet availability concerns, this paper proposes a completely offline, on-device, cross-device, and opensource system to ensure complete data privacy. The proposed system was tested with several datasets, including AI2 Reasoning Challenge, SQuAD 1.1, CoNLL 2003, GSM8K and StrategyQA to evaluate the closed-form question answering (QA), contextual understanding, named entity recognition, mathematical reasoning and truthfulness, respectively, and with five on-device large language models (LLMs), including Gemma3 1B, SmolLM 1.7B, Qwen2 1.5B, TinyLlama-1.1B, and Phi-2. The system achieved the highest score for closed-form accuracy of 1.0. Its performance on reasoning ranged from 0.01 to 0.23. Truthfulness scores ranged from 0.24 to 0.59. High F1 scores for named entity recognition ranged from 0.74 to 0.79, and contextual understanding scores ranged from 0.02 to 0.17 across the different LLMs. The average response time of the system on mobile and desktop devices was evaluated and observed to vary according to system capability and model size. The system allows users to choose between multiple wake words specific to the Indian context. The proposed system functions on limited RAM and in constrained resource environments.
Authors - Roshani Tawale, Jayshri Todase, Manisha Bharati Abstract - Enforcement of helmet regulations and accurate vehicle identification remain essential components of intelligent traffic management systems. Conventional supervision approaches depend heavily on manual inspection, which is labor-intensive and unsuitable for continuous large-scale monitoring. This study presents an automated framework for helmet violation detection and number plate lo-calization using the YOLOv8 deep learning architecture [3]. The proposed system supports static image analysis, recorded video processing, and live-stream detection within a unified pipeline. Performance is assessed using precision, re-call, and mean Average Precision (mAP@50). Experimental findings demonstrate consistent detection reliability and validate the framework’s applicability for real-time traffic surveillance systems.
Authors - Gunjan Pareek, Rajiv Singh, Swati Nigam Abstract - This research examines the transfer learning deep learning models in multimodal human activity recognition based on wearable sensor data. Raw IMU signals are converted to Gramian Angular Field (GAF) images to improve the feature representation and tested on WISDM and PAMAP2 datasets of 18 activity classes. Five CNN models, namely VGG16, MobileNetV2, ResNet50, DenseNet121, and EfficientNetB0, are trained and evaluated in the same conditions and measured by classification accuracy, statistical significance, and computation efficiency. GAF representations are always better than raw signals. DenseNet121 and ResNet50 have 99% accuracy, VGG16 and MobileNetV2 perform competitively and EfficientNetB0 performs worse. Most of the differences in performance are statistically significant (p < 0.05).
Authors - Devang Rupesh Dalvi, Gaurav Suresh Malik, Abhishek Jairaj Kunder Abstract - Prompt engineering has emerged as an essential paradigm in leveraging desired behaviors from large language models (LLMs) without altering their parameters. Although the majority of the current literature has revolved around the introduction of novel prompt engineering strategies, there has been comparatively less emphasis on the contribution of the evaluation and optimization of prompts in concrete systems. In this paper, we offer a specialized review of prompt engineering from an evaluation/optimization centric viewpoint with a larger nod to conceptual developments and illumination rather than detailing the comparisons of approaches. Furthermore, we attempt to establish the concrete importance of prompt engineering via a real-life application, which resulted in improved performances in tasks through the process of prompt refinement and informal evaluations without the need to change the architecture and weights of the models. The paper will also introduce the deficiencies in prompt engineering in the realms of re-producibility, robustness, and the unavailability of standardized approaches in the aspect of concrete evaluations.
Assistant Professor, Assistant Head Research- Department of Information Technology, Vishwakarma Institute of Technology Pune (Affiliated to Savitribai Phule Pune University, Maharashtra, India
Authors - Shruti Thakur, Shilpa Nikhil Bhosale, Priti Prakash Jorvekar, Sandeep Muktinath Chitalkar, Harshala Shingne, Rupali Vairagade Abstract - This study examines the effectiveness of ensemble learning models for detecting fraud in e-wallet transactions under extreme class imbalance and temporal dependence. Using the PaySim bench-mark dataset, a time-aware experimental framework is developed that incorporates forward-chaining evaluation, imbalance-aware resampling, hyperparameter optimisation, probability calibration, and cost-sensitive threshold tuning to reflect real-world deployment conditions. RF and XGBoost are systematically compared across multiple dataset scales and train–test splitting strategies. Empirical findings show that XGBoost consistently outperforms RF, achieving the highest F1-score, maintaining PR-AUC above 0.88, and demonstrating near-perfect ROC-AUC, indicating strong discriminative capability. Following isotonic calibration, XGBoost also produces the lowest Brier score, highlighting superior probability reliability for risk-based decisions. Performance gains plateau beyond a 75% training share, while XGBoost preserves stable performance as the test window expands, unlike RF. Overall, the results support prioritising gradient boosting models, adopting time-aware validation, and integrating calibrated risk scoring in operational e-wallet fraud detection systems.
Authors - Shoh-Jakhon Khamdаmov, Muazzam Akramova, Rano Abdullaevna Sadikova, Azamat Kasimov, Jasurbek Pozilovich Kurbonov, Alisher Bakberganovich Sherov, Dilshoda Akramova Abstract - Attention Deficit Hyperactivity Disorder (ADHD) is one of the most common neurodevelopmental disorders in children, characterized by inattention, hyperactivity, and impulsivity that impair academic and social functioning. Due to its heterogeneous presentation and symptom overlap with other cognitive disorders, early and accurate diagnosis remains challenging. This study proposes a multimodal machine learning framework integrating behavioral, neuroimaging, and physiological data to predict ADHD in children. Convolutional Neural Networks (CNNs) are used to extract features from brain MRI scans, Long Short-Term Memory (LSTM) networks model temporal patterns in physiological signals such as EEG and heart rate variability, and ensemble learning methods incorporate behavioral and clinical attributes. Both feature-level and decision-level fusion strategies are evaluated. Results on benchmark datasets show that the multimodal model consistently outperforms unimodal approaches in accuracy, sensitivity, and F1- score, demonstrating the potential of AI-driven multimodal systems for early, objective, and interpretable ADHD diagnosis.
Authors - Matjere Matsebe, Nobubele Angel Shozi Abstract - It is possible to increase the acceptability of small wind turbines for wind regions with low wind velocities for rural as well as urban sectors by placing them inside diffusers. The research on development of various diffusers is a major re-search area nowadays. Curved flanged diffusers can deliver better performance by adding a cylindrical throat section between converging and diverging sections. This research paper presents a systematic study on short curved flanged diffusers with converging-diverging sections and extended uniform throat between them. Twenty-five diffuser models are studied using Computational Fluid Dynamics using ANSYS Fluent. These models are finalized using the design of experiments for six variables at five levels. The throat diameter for all diffuser models is fixed. The investigation is performed by considering radial average velocity and percentage velocity variation along the radial planes. The global velocities are observed as 1.18 to 1.47 times that of the radial average velocities. The diffuser dimensions are optimized to maximize radial average velocity and to minimize the velocity variation along the radial planes. The diffuser with optimized dimensions is manufactured and tested experimentally in a wind tunnel. Good matching is seen between the predicted results and experimental results. The optimized diffuser has the ability to produce more than two times the power that of the turbine without a diffuser.
Authors - Murodov Gayrat Nekovich, Kholmuhamedov Bakhtiyor Farkhodovich, Avezov Sukhrob Sobirovich, Khudayberganov Nizomaddin Uktambay ogli, Yunusova Maftuna Shokirovna, Mansurova Shahinabonu Najmiddin qizi Abstract - The classification of ECG signals continues to be a major focus in intelligent healthcare systems, especially for the early identification of cardiac arrhythmias. In this work, we propose a hybrid probabilistic neural strategy that integrates Bayesian Networks with Artificial Neural Networks (ANNs) to enhance the reliability of ECG classification. The approach begins by extracting informative ECG features, such as crosscorrelation and phase-based characteristics. A Bayesian Network is then applied to model the probabilistic dependencies among these features and identify those most relevant to classification. At the same time, an ANN is trained on the refined feature set to learn complex non-linear patterns present in the signals. The two models are subsequently combined through a weighted voting mechanism to form an ensemble classifier. Experimental evaluation using an ECG dataset indicates that the proposed ensemble achieves higher accuracy and stability compared to its individual components. Notably, the method demonstrates strong capability in distinguishing multiple arrhythmia categories, which are typically difficult to classify. Overall, the results highlight the promise of hybrid probabilistic–neural models for improving automated ECG interpretation and supporting more accurate diagnosis of cardiac abnormalities.
Authors - K S Shubham, Uma Mudengudi, Ujwala Patil Abstract - Secure, compliant, and interoperable data sharing remains a core bottleneck for cross-organizational analytics and AI, particularly under evolving privacy regulations, contractual obligations, and adversarial threats. This paper introduces HARMONIA, a pluggable, risk-aware data sharing framework that integrates policy-as-code enforcement, continuous compliance monitoring, provenance-grade evidence, and revocation with machine unlearning. HARMONIA is inspired by the iterative Analyzer–Mechanic and Conductor–Observer operational pattern described in the HARMONIA strategic perspective, generalizing its quality-gate-and-repair loop to a policyand- risk-gated release lifecycle. We formalize an architecture that separates governance, control, and data planes; define a release-mode lattice that enables explainable fallbacks among raw export, masking, kanonymity, differential privacy, synthetic data, query-only access, and federated compute; and propose an evidence model aligned with W3C PROV. We provide a proof-of-concept (POC) blueprint implemented with commodity components (OPA, OAuth2/OIDC, PostgreSQL, and object storage) and specify interfaces that support end-to-end request-to-release-to-revocation workflows, including batch-scoped unlearning for model derivatives. The paper concludes with an evaluation methodology and a standards-aligned roadmap for deployment in sovereign data spaces.
Authors - Mehzabul Hoque Nahid, Fatema Tuz Zahra, Mubashshir Bin Mahbub, Saleh Ahmed Jalal Siam Abstract - Personalizing learning in higher education presents a significant challenge due to the difficulty of providing individual feedback to large student cohorts. This study proposes an intelligent tutoring system based on a multi-agent architecture utilizing Large Language Models (LLMs) to address scalability and adaptability issues. The proposed architecture integrates two complementary subsystems: a reactive module that answers student queries using Retrieval-Augmented Generation (RAG) to ensure accuracy based on course materials, and a proactive module that autonomously analyzes student profiles to generate personalized study plans without direct instructor intervention. The system was implemented using Lang- Graph for agent orchestration and MongoDB for state persistence. Experimental validation was conducted using a curated golden dataset from a university course. Results demonstrate a retrieval precision of 94.2% and a faithfulness score of 87.8%, significantly mitigating hallucinations common in monolithic models. Furthermore, the operational cost analysis indicates high financial viability for mass implementation. This dual approach offers a robust solution for automated, highquality educational support, effectively bridging the gap between standardized teaching and personalized learning needs.
Authors - Vemuri Bharath Kumar, Anjan Babu G Abstract - Healthcare data scarcity poses significant challenges for machine learning applications in clinical settings, particularly for conditions with limited patient populations. This paper presents a novel quantumenhanced data augmentation framework that addresses this challenge through a three-pillar architecture: Quantum Random Number Generation (QRNG) for true randomness, Statistical AI for intelligent parameter optimization, and Generative AI for clinical interpretability. Our implementation utilizes Bell state quantum circuits to generate genuinely random perturbations, ensuring higher entropy than classical pseudorandom methods. The framework incorporates medical domain knowledge through constraint-aware augmentation, maintaining clinical validity while generating synthetic patient records. Experimental evaluation on the Pima Indians Diabetes dataset (768 samples, 8 features) demonstrates that our quantum-enhanced approach achieves 100% medical constraint compliance while generating high-quality synthetic data. The system provides both command-line and web interfaces, with automatic fallback to classical methods when quantum resources are unavailable. Our contributions include: the first practical application of quantum computing to healthcare data augmentation, an AI-driven optimization system that automatically determines augmentation parameters, integration with large language models for non-technical summarization of validation reports, and a production-ready implementation with comprehensive validation mechanisms. The framework represents a significant advancement in synthetic medical data generation, offering a scalable solution for addressing data scarcity in healthcare AI applications.
Authors - Sandhya Awate, Vipin Kumar Gupta Abstract - Rural communities face significant challenges in accessing essential healthcare services due to language barriers, limited health literacy, and insufficient medical support. Difficulties in understanding medical information, communicating symptoms, and interpreting diagnostic reports further hinder effective healthcare delivery. Additionally, unreliable internet connectivity restricts the reach of conventional digital health platforms. To address these challenges, this paper presents a Multilingual AI Health Assistant designed to operate on low-cost edge devices, enabling offline functionality to ensure continuous access and data privacy in low-connectivity areas. The proposed system integrates Artificial Intelligence (AI), Machine Learning (ML), Natural Language Processing (NLP), Optical Character Recognition (OCR), and speech recognition, allowing users to interact in their native languages via text or voice. It analyzes user-reported symptoms to predict probable health conditions, translates complex medical reports and prescriptions into simplified, localized explanations, and provides recommendations for nearby healthcare facilities. Unlike internetdependent telemedicine systems, this edge-based solution processes data directly on the device, safeguarding sensitive health information while maintaining reliability. By bridging linguistic and literacy gaps, the proposed assistant empowers rural populations with accessible and actionable healthcare insights, ultimately improving health outcomes in underserved regions.
Authors - P. Sivaperumal, R. Naresh, S. Prawin, B. E. Viruthatchanan Abstract - The food portion estimation is a critical component of automated dietary assessment systems, enabling better monitoring of nutritional intake and supporting healthcare, weight management, and public health applications. Traditional self-reporting methods are often inaccurate and time-consuming, motivating the need for computer vision–based approaches that can reliably estimate food portions from images captured in real-world conditions. This paper presents deep learning pipeline for food portion estimation that integrates image preprocessing, deep learning–based segmentation, and geometric volume computation. The data preprocessing with Mask R-CNN used for precise food seg-mentation, providing pixel-level masks and bounding boxes that isolate individual food items from complex backgrounds. The segmented mask is used to estimate the pixel area of the food region. Experimental evaluation demonstrates that the proposed method achieves high segmentation accuracy, with a segmentation IoU of 87.6%, precision of 90.3%, recall of 88.9%, and an F1-score of 89.6%. The pixel area estimation error is limited to 6.8%, resulting in an overall portion estimation accuracy of 89.1%, indicating reliable and consistent performance across different food images. The proposed framework highlights the effectiveness of combining deep instance segmentation with geometric volume estimation for accurate food portion assessment. Future work will focus on multi-view image integration and real-time deployment in mobile dietary monitoring systems to enhance robustness and scalability.
Authors - Chalani Dinitha, Saadh Jawwadh Abstract - Automated Image Enhancement from CCTV surveillance relies heavily on accurate image segmentation; however, real-world footage is often degraded by low illumination, motion blur, occlusion, and background clutter, causing conventional segmentation models to lose boundary precision and small object details. This paper proposes EdgeLite-CrimSegNet, a novel lightweight boundary-aware segmentation network designed specifically for crime scene analysis. Unlike existing fast segmentation models that prioritize global context, the proposed architecture adopts a boundary-first learning strategy, where crime-relevant contours are explicitly extracted and refined before region-level segmentation. A compact edge-aware encoder, boundary-guided feature refinement module, and progressive region filling strategy are introduced to improve segmentation accuracy while maintaining real-time performance. Experiments on CCTV frames derived from the UCF-Crime dataset demonstrate improved boundary preservation, higher IOU, and better segmentation of overlapping and small objects compared to conventional lightweight segmentation networks, confirming the suitability of EdgeLite-CrimSegNet for real-time surveillance applications.
Assistant Professor, Assistant Head Research- Department of Information Technology, Vishwakarma Institute of Technology Pune (Affiliated to Savitribai Phule Pune University, Maharashtra, India
Authors - Viet Anh DUONG, Hai Phong BUI, Van Son NGUYEN Abstract - This article presents a neuro-symbolic modelling approach grounded in qualitative data collected from 25 sports clubs located in R´eunion. The study develops a methodological chain linking structured semantic extraction, ontological formalisation in OWL, and agent-based simulation implemented in NetLogo. Rather than modifying structural scenarios across experiments, the design introduces two contrasting organisational sensitivity profiles derived from field observations: a damped profile and a high-gain profile. The structural configurations remain identical between profiles; only the coefficients of the commitment update function vary, ensuring strict experimental comparability. Results indicate that identical structural conditions produce differentiated collective trajectories depending on internal sensitivity parameters. In highgain configurations, dominance-weighted interactions increase variance and generate polarised engagement distributions, whereas damped configurations maintain relative stability across scenarios. These findings suggest that modelling organisational sensitivity parameters is critical for understanding the robustness of digitally mediated collaboration in volunteer-based organisations.
Authors - Allezandra A. Adriano, Joshua Basile Mhar L. Austria, Benjamin L. Carnate, Xamantha Angelique E. Ruiz, Wilben Christie R. Pagtaconan Abstract - Plant diseases due to various pathogens can cause significant loss in yield and productivity. The classification of these diseases is necessary to prevent damage to crops. For classification, a large number of Machine learning and deep learning algorithms have been developed. In this research, five classes of plant leaves and a further fifteen different diseases of these plants (three subcategories for each class) are used for classification. In the proposed methodology, we have used three pre-trained models, namely, ResNet 152v2, InceptionResNetV2, and mGoogleNet, and a custom-built model. This research has used three basic steps to classify the disease categories, namely image preprocessing, image segmentation, and feature extraction. Fifteen thousand plant leaf images have been collect-ed from the online available Kaggle PlantVillage dataset. This data is present in a JPG file format. After the class label distribution of the dataset, the dataset is first trained and then tested on these deep learning models. The label distribution is done in such a way that each of these fifteen categories has 80% training images and 20% validation images. We have used different performance measures, namely, precision, recall, F1-score, and support, to calculate the accuracy. The obtained validation accuracy of ResNet152V2 is 97%, GoogleNet is 96%, Incep-tionResNetV2 is 93%, and a custom-built model is 99%. These results show that the custom-built model has attained the highest accuracy. These models can also be used to build a recommender system framework for the recommendation of fertilizers in the future.
Authors - E. Praveen Kumar, Shankar Lingam. M Abstract - Quantum computers are a major threat to the existing encryption mechanisms. In terms of security, the traditional encryption algorithm depends on complex problems like discrete logarithm as well as factorization of integer. Shor’s algorithm is believed to break the current Public Key Encryption algorithms such as Advanced Encryption Standard (AES). Therefore, several research are carried out in the area of PQC (Post Quantum Cryptography). PQC are based on very complex mathematical problems like Learning with error (LWE) which are robust against quantum computers. The National Institute of Standard and Technology (NIST) has initiated several rounds of standardization process for PQC algorithms, among which NTRU, SABER, CRYSTAL-KYBER are the leading candidates. CRYSTALS-KYBER (Kyber) is the first chosen PQC for standardization. This works explores the recent development in Crystals Kyber implementation and its optimization. Researchers can approach for new research challenges and target for improvement thereby increasing efficiency.
Authors - An Doan Van, Dong Nguyen Doan, Quynh Tran Duc, Thuan Nguyen Quang, Bao Phan Gia, Hieu Doan Minh, Van Khanh Doan Abstract - Performance bottlenecks in Python programs arise from a wide variety of sources, and no single technique reliably catches them all. This paper proposes CodeForge, a sequential three-stage optimization system that unites deterministic Abstract Syntax Tree (AST) inspection, CodeBERT embedding-based retrieval, and Gemini LLM-driven rewriting into one end-to-end pipeline. A rule engine in the first stage pinpoints well-known structural problems; a neural similarity search in the second stage captures harder-to-spot variants; and a Gemini LLM in the third stage performs the actual rewrite, guided by a structured hint block assembled from both preceding stages. Before any result is returned, a configurable validator rejects changes that fail minimum speedup, memory, or complexity criteria. Alongside each accepted optimization, a composite confidence score and a plain-language rationale are produced. Tests on six representative Python patterns show that hint-guided LLM prompting raises successful detection from four to six out of six cases compared with unguided prompting, while the validation layer blocks every harmful transformation in the test suite. The system is available as a FastAPI REST service accepting both raw source text and uploaded .py files.
Authors - Jyotika R. Yadav, Arpit A. Jain Abstract - Internet of Things (IoT) with AI techniques help healthcare industry for patient monitoring and diagnosis. Wearable devices integrated with the Internet of Medical Things (IoMT) have transformed modern healthcare by enabling continuous, real-time monitoring of physiological parameters. The rapid evolution of Artificial Intelligence (AI), Machine Learning (ML), Deep Learning (DL), edge computing, and federated learning has further enhanced the reliability, privacy, and intelligence of such systems. Wearable devices like smart watch or smart sensors help doctors to monitor patient’s daily activities. However, these devices generate huge amount of data on day-to-day basis which makes analysis, monitoring, and diagnosis challenging. Machine Learning or Deep Learning models used for handling such large healthcare data. This survey consolidates and critically reviews recent research works to provide a holistic understanding of the current state-of-the-art in wearable AI-enabled healthcare. A detailed comparative analysis is provided to highlight similarities, differences, strengths, and limitations of existing approaches. Finally, key challenges and future research directions are discussed to guide the development of secure, scalable, and intelligent wearable healthcare solutions.
Authors - Shweta H. Jambukia, Pooja R. Makawana, Prapti G. Trivedi Abstract - This paper presents a case study on a High Voltage Jet (HVJ) electric boiler, focusing on current unbalance (CU) risk identification and mitigation us ing a combined data-analytics and Failure Mode and Effects Analysis (FMEA) framework. Power-quality assessment follows IEC 61000-4-30 for voltage un balance (VU), while CU interpretation refers to NEMA MG-1 and IEEE recom mendations. The proposed workflow integrates (i) instrument classification (Class A for voltage), (ii) time synchronization across logger/PLC/power-quality analyzer to avoid timestamp drift, and (iii) historian-based data pre-processing (outlier cleaning, scaling, and missing-data handling) prior to statistical analysis. Results show an average CU of 6.85% with a standard deviation of 0.48% and a maximum of 15.92%, indicating operational periods exceeding common industry limits. FMEA highlights electrode aging/damage, loose/corroded cable connec tions, and supply power-quality issues as the dominant contributors. Recom mended actions include online phase-current monitoring, improved water-chem istry and blowdown management, and control optimization of the VFD-driven boiler circulation pump (BCP).
Authors - Priyanka K, Vinay R K, Vansh Jain, Vinit Kulkarni Abstract - This study examines the influence of both demographic and natural factors on climate change risk perception in New Zealand. Using data from a nationally representative survey, the analysis applies exploratory factor analysis to construct a composite measure of risk perception, followed by correlation and regression modeling to evaluate the relative contribution of environmental exposure and human characteristics. The findings indicate that while natural factors such as temperature anomalies and extreme weather exposure significantly shape perceived risk, demographic variables including prior disaster experience, trust in scientific institutions, and media exposure exert a stronger overall influence. These results underscore the importance of incorporating social and behavioral dimensions into climate risk assessments and policy development to enhance public engagement and adaptive capacity.
Authors - Piyush Tewari, Rohit, Rujal Agarwal, Yanshi Sharma Abstract - Current Network Intrusion Detection Systems (NIDS) typically analyze traffic as independent tabular records, largely ignoring the relational and temporal dependencies inherent in real-world communications. This limitation is particularly critical for detecting botnets, which rely on coordinated, evolving interactions rather than isolated malicious packets. To address this, we propose a topology-aware framework that models network traffic as a sequence of dynamic communication graphs. Using the CICIDS2017 dataset, we construct sliding-window snapshots where IP addresses form nodes and flows form edges. A spatiotemporal graph neural network is employed to learn evolving structural representations, integrated with a novel learnable gated fusion mechanism that adaptively balances graph-based context with conventional flowlevel statistics. The model is optimized using a hybrid objective combining class-weighted cross-entropy and center loss to mitigate data imbalance. Experimental results demonstrate that the framework achieves improved performance on structural attacks, with botnet detection reaching an AUC of 0.999. Furthermore, the learned gating values reveal a strong model preference for topological features over static statistics, empirically validating that structural context is superior for identifying coordinated threats. These findings underscore the effectiveness of spatiotemporal modeling in enhancing the robustness and interpretability of next-generation NIDS.
Authors - Bikkam Hemanth Reddy, Allu Eswar Kaushik, Tiyyagura Mohit Reddy, Kuruboor Venkatesha Deepak, Bharathi D Abstract - Cloud cover generally limits the applicability of optical remote sensing images for tasks such as agriculture monitoring and disaster relief. Cloud removal is an inherently difficult problem because of the lack of spatial structures and spectral information. To effectively remove cloud contamination from SAR and optical images, we propose a speckle-aware global cross-attention network. The proposed SAR-optical cloud removal network architecture consists of a dual encoder with a global cross-attention mechanism that allows for effective cross-modal interactions. Additionally, a refining module and symmetric decoders improve the accuracy of the reconstructed image. Furthermore, we propose a speckle-aware gating mechanism to perform speckle filter adaptation. The experimental results affirm that our proposed network outperformed the baseline by increasing Peak Signal-to-Noise Ratio(PSNR) by +0.86 dB, Structural Similarity Index Measure(SSIM) by +0.142, and reducing the spectral distortion of the image. Additionally, we noticed a decrease in the Root Mean Square Error(RMSE) and Spectral Angle Mapper(SAM) values. This infers that selective SAR-Optical fusion with an adaptive noise-aware gating mechanism improves the accuracy of cloud-free optical images and optical remote sensing images.
Authors - Azamat Kasimov, Kholida Bekpolatovna Saidrasulova, Zebo Abduxalilovna Shomirova, Shoh-Jakhon Khamdаmov, Safiya Karimova, Dilshoda Akramova, Doniyor Niyozmetov Abstract - Inconsistent medication intake is a major issue, especially for elderly individuals and patients with memory problems [1]. The MediMitra: Voice Enabled Medicine Alert System seeks to tackle this problem by offering an automated, low-cost and user-friendly medication reminder solution. The system combines Raspberry Pi with Optical Character Recognition (OCR) technology to pull medicine names, dosage details and intake times directly from scanned prescriptions. This reduces manual input and user reliance. The information is stored in a central database and connected to a scheduler that sends timely voice alerts through smart speakers or Bluetooth devices. This ensures users receive reliable and easy-to-access reminders. The OCR module is designed for high accuracy in processing printed prescription images by using image preprocessing techniques like noise reduction and thresholding, which helps in effectively extracting key medication details [2]. The system focuses on accessibility, affordability and ease of use in home or clinical settings. Overall, MediMitra provides a useful technological solution to improve medication adherence and supports independent living. It also has potential for future integration with health-monitoring systems.
