Volume 14 | Issue 8 |

 Abstract

Online examinations and remote coding assessments require proctoring systems that can verify candidate authenticity, detect suspicious behaviour, preserve evidence and support fair human review. This paper presents a multimodal AI-based online proctoring system designed for secure coding assessments. The proposed system integrates captcha-less behavioural login analysis, OTP verification, a browser-based coding interface, MTCNN face analysis, MediaPipe FaceMesh gaze estimation, YOLOv8 object detection, audio activity monitoring, cooldown-based violation aggregation and automated integrity- report generation. Unlike conventional webcam-only monitoring, the system represents a candidate session as a multimodal stream and applies mathematical decision rules to authentication, visual attention, object presence, audio activity and event acceptance. The implementation is analysed using project source code and generated violation logs. A case-study session containing 100 accepted events demonstrates the complete monitoring pipeline and shows that gaze deviation, face absence, multiple-face presence and prohibited-object events can be captured in a structured evidence model. The paper further discusses computational complexity, deployment security, privacy, limitations and validation requirements for institutional use.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

online proctoring, AI proctoring, coding assessment, computer vision, MTCNN, MediaPipe, YOLOv8, gaze tracking, academic integrity.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Alzheimer's disease (AD) is a progressive neurodegenerative disorder in which early and accurate prediction of disease progression remains a clinical challenge. This study proposes an attention- based multimodal deep learning framework that integrates structural magnetic resonance imaging (MRI) and clinical features for multi-class progression prediction. MRI volumes are processed using a three- dimensional convolutional neural network to extract spatial biomarkers, while demographic and cognitive variables are encoded through a clinical feature encoder. A modality-level attention fusion mechanism is introduced to adaptively weight imaging and non-imaging information, enabling the model to learn clinically meaningful representations. The system is trained and evaluated on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset with class imbalance handled through weighted sampling and cost-sensitive learning. The framework also supports explainable artificial intelligence using gradient- based localization and feature attribution for clinical transparency. Experimental results demonstrate improved macro-F1 performance compared with unimodal approaches and provide interpretable attention distributions indicating modality contribution.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Alzheimer's Disease Prediction, Multimodal Deep Learning, MRI Clinical Fusion, Explainable Artificial Intelligence, ADNI Dataset

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Computing advancements have changed the nature of cybersecurity dramatically, driving a move from passive rules-based detection towards a data-driven approach using intelligent cyber defense systems. Cybersecurity data science is a fundamental aspect that can bridge the gap between raw security data and intelligent threat intelligence. This paper offers a comprehensive overview of cybersecurity data science by discussing structured collection and processing of security data from a variety of data sources (network traffic logs, intrusion detection systems, endpoint telemetry etc.), its utilization of machine learning, deep learning, and behavioral analysis in identifying significant patterns and extracting actionable information to achieve real time threat detection, vulnerability analysis, and automated incident response. As a core aspect, it proposed a novel machine learning based multi-layered framework that aims to aid in automated intelligent decisions for a variety of cyber-threats. It concludes by summarizing crucial research issues including adversarial machine learning, data imbalance and model interpretability through authoritative literature, along with suggestions for future developments.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Machine Learning Approaches in Cybersecurity Data Science: An Analytical Overview

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The rapid growth of social media platforms has significantly increased the spread of online scams, phishing attacks, fake advertisements, fraudulent investment schemes, and impersonation attacks. Traditional scam detection systems primarily focus on textual analysis and fail to identify complex multimodal scams that combine text, images, videos, and malicious URLs. This research proposes a Multi-Modal Scam Detection System using Explainable Artificial Intelligence (XAI) to improve the accuracy, transparency, and reliability of scam detection in social media environments. The proposed framework integrates Natural Language Processing (NLP), image feature extraction, metadata analysis, and machine learning techniques for identifying fraudulent content. The system utilizes transformer-based text classification, Convolutional Neural Networks (CNN) for image analysis, and ensemble learning methods for multimodal fusion. Explainable AI techniques such as SHAP (SHapley Additive Explanations) and LIME (Local Interpretable Model-Agnostic Explanations) are incorporated to provide interpretable predictions and increase user trust. Experimental results demonstrate that the proposed model achieves higher accuracy, precision, recall, and F1-score compared with traditional machine learning methods. The research contributes toward secure digital communication and reliable social media monitoring systems.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Multi-Modal Learning, Scam Detection, Social Media Security, Explainable AI, Deep Learning, Cybersecurity, NLP, CNN, XAI.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Organisations create a lot of unstructured data, like policies, manuals, and reports. Traditional keyword-based search doesn't capture semantic meaning and context, which makes it hard to find what you're looking for. Even though Large Language Models (LLMs) can understand natural language, they often give wrong or made-up answers when used alone, especially when it comes to new or domain-specific information. This paper presents an Enterprise Retrieval-Augmented Generation (RAG) system that integrates semantic retrieval through dense vector embeddings with context-aware LLM generation. The system handles documents in many formats, uses FAISS to index them, and then uses a GPT-4 backend to give accurate answers that come from the right source. It has a dual-mode architecture for strict retrieval and better explanations, as well as a backup system for queries that aren't very relevant. Tests on a dataset of 500 documents show that it works well, with an 88% retrieval rate,improved response time (1.2s vs. 1.9s baseline), and support for multiple file formats, making it suitable for auditable, enterprise environments.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Retrieval-Augmented Generation; Enterprise Knowledge Management; Semantic Search; Large Language Models; Dense Vector Embeddings; Hallucination Mitigation; FAISS; Dual-Mode Architecture; Response Explainability

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Modern professional environments increasingly rely on virtual meetings, yet extracting action-able insights from lengthy discussions remains a persistent challenge. Participant's often miss critical details, action items, or decisions during fast-moving conversations. Existing solutions typically rely on cloud- based transcription services, which introduce latency, privacy concerns, and dependency on external infrastructure. This paper presents an edge-based agentic meeting assistant architecture capable of performing real-time transcription, contextual understanding, and automated task execution during live meetings. The proposed system integrates a streaming speech recognition pipeline, a locally deployed large language model (LLM), and an agent orchestration framework to analyze conversations and trigger intelligent actions including generating summaries, sending emails, and performing web searches. The architecture emphasizes low latency, enhanced privacy, and modular scalability by executing most processing locally on user devices. Experimental deployment confirms that the system delivers accurate live transcripts and intelligent meeting assistance while minimizing reliance on external cloud services. A detailed performance comparison with cloud-based alter-natives demonstrates the practical advantages of the edge-first approach.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Edge Computing, Agentic AI, Large Language Models, Meeting Intelligence, Real-Time Transcription, VOSK, Llama, Ollama, Speech Recognition, Task Automation

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The rapid evolution of autonomous vehicle technology demands intelligent systems capable of understanding and interpreting complex traffic environments in real time. This paper presents an Automated Traffic Situation Analysis System designed to enhance the perception and decision-making capabilities of autonomous vehicles. The proposed system leverages advanced deep learning techniques, including Convolutional Neural Networks (CNNs) and real-time object detection models such as YOLO, to identify and classify key traffic elements such as vehicles, pedestrians, traffic signals, and road conditions. Furthermore, the system incorporates temporal and spatial analysis to evaluate traffic density, detect anomalies, and predict potential hazards. A scalable and low-latency processing pipeline ensures efficient performance in dynamic and high-density traffic scenarios. Experimental results demonstrate that the system achieves high accuracy and reliability, making it suitable for real-world deployment. By integrating perception, analysis, and prediction, the proposed framework significantly improves safety, efficiency, and adaptability in autonomous driving systems.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

AUTOMATED TRAFFIC SITUATION ANALYSIS SYSTEM FOR AUTONOMOUS VEHICLES USING NOVEL HYBRID ML MODEL

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Urban navigation in megacities such as Chennai, India, presents multifaceted challenges arising from high traffic density, seasonal flooding, accident-prone zones, and the inadequacy of existing systems that optimize solely for travel time and distance. This paper proposes a comprehensive AI-powered Smart Navigation System (SNS) for urban route optimization that integrates multi-objective routing, machine learning-based risk prediction, and real-time data fusion. The proposed architecture combines graph-theoretic routing on the OpenStreetMap road network via OSMnx and NetworkX, an XGBoost-based accident risk classifier, a time-series congestion predictor, and a flood-zone awareness layer specifically calibrated for Chennai's monsoon environment. The system computes four distinct route classes--fastest, shortest, safest, and traffic-optimized -- scored by a composite safety metric: Safety = 100 - (AccidentRisk + FloodRisk + CongestionPenalty). Evaluation on a real-world Chennai road network demonstrates a route safety improvement of 31.4% over shortest-path baselines, an 18.7% reduction in average travel time under congested conditions compared to static routing, and a flood-zone avoidance accuracy of 94.2%. A full multi-modal transportation planner integrating metro, bus, and driving modes is additionally provided. Results confirm the proposed SNS significantly outperforms conventional navigation approaches and offers a scalable framework for other complex urban environments.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Urban route optimization, intelligent navigation, multi-objective routing, accident risk prediction, flood-aware navigation, machine learning, graph-based pathfinding, XGBoost, OpenStreetMap, Chennai, ITS.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This research paper explores the themes of sexual desire, bodily autonomy, and socio- political defiance through the character of Subbamma in Kuvempu's landmark Kannada novel, Kanooru Subbamma Heggadithi (1936), translated into English as The House of Kanooru. Situated within a rigidly patriarchal and feudal framework, the study examines how Subbamma--married to Chandrayya Gowda, a tyrannical village headman significantly older than her--embodies the volatile tension between institutional marital coercion and the untamed desires of youth. Grounded in a textual analysis of her initial romantic yearning for the younger Hoovayya, her subsequent disillusionment with marriage, and her ultimate assertive rebellion, this paper argues that Subbamma's desire functions as more than an internal psychological state. Instead, it operates as a subversive socio-political force that directly challenges the hegemony of the Kanooru household. Ultimately, this study demonstrates how Kuvempu constructs female desire not as a passive, suppressed emotion, but as a site of profound bodily struggle, agency, and anti-patriarchal resistance.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Kuvempu, Kanooru, Subbamma, Desire, Patriarchal Oppression, Feudalism, Resistance

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Clay (earthenware/terracotta) cookware has been used across cultures for centuries for cooking, storing water, fermenting foods and serving traditional dishes. Its porous aluminosilicate matrix, relatively low thermal conductivity, moisture-retaining behavior and gradual heat transfer give it distinctive culinary properties. This review examines the material composition, physicochemical and thermal characteristics, nutrient and mineral transfer, food-safety considerations, potential health implications, cultural significance, environmental advantages and socioeconomic relevance of clay cookware. Available evidence suggests that clay vessels can provide slow and relatively uniform cooking, moisture retention and desirable sensory characteristics. However, mineral migration depends strongly on clay composition, firing conditions, food acidity, temperature and repeated use. Poorly manufactured or glazed pottery may also present a risk of contaminant migration, particularly heavy metals. Claims concerning prevention or treatment of chronic diseases should therefore be interpreted cautiously because robust clinical evidence remains limited. Clay cookware offers a potentially sustainable alternative to some modern cookware, but standardization of raw materials, firing, glazing, contaminant testing and performance is needed before broad health claims can be made. Future research should focus on controlled comparative studies of nutrient retention, contaminant migration, sensory quality, energy use and long-term consumer safety.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

clay cookware; earthenware; terracotta; nutrient retention; mineral leaching; food safety; thermal properties; sustainability; traditional cookware

  License

Creative Commons Attribution 4.0 and The Open Definition