Authors: S. Balaji, N. Poyyamozhi

Abstract: Currently, the field of project management faces increasing uncertainty as projects must deal with changing requirements, resource shortages, and the unpredictable effects of human actions, technical systems, and external events. However, existing data-driven models have failed to provide interpretable results, preventing project managers from identifying the factors that create risks. Thus, this research presents a lightweight and explainable data-driven decision support system that enables project risk prediction and risk management in complex project management environments. The devised methodology employs a Project Management Risk Dataset, which includes project demographics and operational metrics, human factors, organizational context, technical aspects, and external influences. Moreover, a comprehensive data reliability testing is conducted through pre-processing methods for categorical attributes, one-hot encoding, and Min-Max normalization of budget and timeline, and risk metrics. Advanced feature engineering uses graph-based feature relationships to identify hidden project attribute dependencies, Graph Signal Processing to create project attribute dependencies, and LASSO with polynomial feature expansion to achieve optimal results. The proposed TAM-Lite architecture integrates TabNet, a mini autoencoder, and a shallow multilayer for project risk prediction. Moreover, stage-wise training is conducted based on Gradient Boosted Rule Sets with Extreme Learning Machines and fuzzy logic classification. The model generates risk level probabilities, which are evaluated through Bayesian Networks and counterfactual explanations to deliver clear and actionable risk reduction recommendations.

DOI: https://doi.org/10.5281/zenodo.19483001

 

Published by: vikaspatanker

Authors: Dr. Satish Kumar

 

Abstract: This study investigates the unsteady magneto hydrodynamic free convective flow of a viscous, incompressible, and electrically conducting fluid past an infinite vertical porous plate with porous medium and applied uniform magnetic field in the direction of the flow. The effect of injection/suction velocity and the magnetic field on the flow field, skin friction and heat transfer are reported and discussed in detail. The Hartmann number and porosity parameter influence the flow velocity, while the Prandtl and Grashof number govern the heat transfer characteristics. The governing partial differential equations for momentum and energy are transformed into a dimensionless form using appropriate similarity variables.

DOI: https://doi.org/10.5281/zenodo.19482335

 

Published by: vikaspatanker

Authors: Mr. Durgunala Ranjith, Gaddam Abhinay Reddy, Guda Raja Krishna, Tejavath Nithin Nayak

Abstract: The “Smart and Intelligent Web Traffic Analytics and Monitoring System” is designed to track and analyze website traffic in a simple and effective way. It collects real-time data about users, page visits, and browsing behavior. The system helps website administrators understand how users interact with their website. It can identify traffic patterns and detect unusual or suspicious activities. Visual reports and dashboards make the data easy to read and interpret. This system supports better decision-making to improve website performance and security.

 

Published by: vikaspatanker

Authors: Mrs. D. Srilatha, Shaik Umaiza Bhanu, Yanamala Lalith, Shaik Amin Sadik, Kamunuri Kasi Ganesh

Abstract: — In the digital era, managing lost and found items efficiently remains a challenge due to reliance on manual methods and unstructured reporting systems. Traditional approaches such as notice boards and text- based communication often result in disorganized data, delayed responses, and low matching accuracy. These limitations highlight the need for an intelligent and automated solution. This paper presents SMARTLOFO: AI Powered Lost and Found Platform, a full- stack web application designed to streamline the process of reporting, tracking, and retrieving lost items. The system is developed using React for the frontend and a Python-based FastAPI backend, with MongoDB/SQLite for data storage. It provides a user-friendly interface along with secure authentication using JWT and bcrypt. A key contribution of the system is the integration of an AI-powered smart matching algorithm. Using Google Gemini, the system performs image analysis to extract item descriptions, categories, and features. These attributes are processed using a scoring-based matching mechanism that evaluates similarity based on category, extracted features, location, and time proximity. Matches exceeding a defined threshold are automatically identified, and users are notified via an email notification system. The platform is deployed on a cloud environment, enabling real-time interaction and accessibility. Despite its advantages, the system depends on user participation and input accuracy. Future enhancements include improving scalability and incorporating advanced machine learning models. Overall, SMARTLOFO demonstrates an intelligent and scalable approach to modernizing lost-and-found systems using artificial intelligence and full-stack technologies.

DOI:

 

 

Published by: vikaspatanker

Authors: Shailja Jha

Abstract: The rapid proliferation of Artificial Intelligence (AI) technologies has exposed significant limitations in traditional state-centric regulatory frameworks, particularly in complex and diverse jurisdictions such as India. This paper advances the concept of a transition from a “regulatory state” to a “regulatory space,” emphasizing the distributed, multi-actor nature of AI governance. Drawing on comparative regulatory theory, the study analyzes how India’s AI governance is characterized by institutional fragmentation, overlapping mandates, and sector-specific regulatory interventions rather than a unified legal framework. By examining key regulatory bodies, policy instruments, and emerging guidelines across domains such as data protection, digital markets, and sectoral compliance, the paper maps the contours of India’s evolving AI governance ecosystem. It further compares India’s approach with global models, including the European Union’s risk-based regulatory regime and the United States’ market-driven governance structure, to highlight divergences and convergences in regulatory philosophy. The analysis demonstrates that India’s fragmented governance structure, while often viewed as a limitation, may also function as a flexible “regulatory space” that enables adaptive, context-sensitive oversight. However, this flexibility comes with challenges related to coordination, accountability, and enforcement consistency. The paper concludes by proposing a hybrid governance model that integrates centralized policy direction with decentralized regulatory innovation, thereby aligning India’s AI governance with both domestic priorities and global regulatory trends.

DOI: https://doi.org/10.5281/zenodo.19481225

Published by: vikaspatanker

Authors: P. Saranya, Ms. E. Sheela

Abstract: The rapid expansion of digital payment platforms has significantly transformed financial transactions worldwide. In India, the Unified Payments Interface (UPI) has emerged as one of the most widely adopted real-time payment systems due to its speed, convenience, and low transaction cost. However, the increasing popularity of UPI has also led to a substantial rise in fraudulent activities, including phishing attacks, unauthorized fund transfers, identity theft, and account takeover incidents. Traditional rule-based fraud detection systems rely on static thresholds and predefined heuristics, which are often unable to adapt to evolving fraud patterns and complex transaction behaviors. Furthermore, fraud detection datasets are typically highly imbalanced, where fraudulent transactions represent only a small fraction of the total data, making accurate detection more challenging. To address these limitations, this study proposes FraudShield-UPI, a machine learning-based fraud detection framework designed to improve the accuracy and reliability of fraud identification in digital payment systems. The proposed framework integrates Synthetic Minority Oversampling Technique (SMOTE) to handle class imbalance, Principal Component Analysis (PCA) for dimensionality reduction, and Extreme Gradient Boosting (XGBoost) for high-performance classification of fraudulent transactions. The system is implemented as a web- based application using the Flask framework, enabling real-time fraud prediction and interactive transaction analysis. In addition to the proposed model, a comparative evaluation platform is developed to benchmark traditional machine learning algorithms including Decision Tree, Support Vector Machine (SVM), and Random Forest using the same dataset and evaluation metrics. Experimental evaluation on a simulated UPI transaction dataset demonstrates that the proposed SMOTE-PCA-XGBoost model significantly outperforms baseline models in terms of accuracy, precision, recall, and F1-score, while effectively reducing both false positives and false negatives. The results highlight the capability of the proposed framework to detect fraudulent transaction patterns with improved reliability. The modular architecture and web-based deployment further demonstrate the practical feasibility of integrating the system into real- world financial platforms for enhanced digital payment security.

DOI: https://doi.org/10.5281/zenodo.19483516

 

Published by: vikaspatanker

Authors: Dr. C. Saravanabhavan, Akhil R

Abstract: Recent advances in deep learning have en-abled highly realistic synthetic speech, creating serious risks such as impersonation, fraud, and misuse of voice-based authentication systems. Detecting AI-generated speech is increasingly difficult because modern text-to-speech and voice conversion models can closely imitate human prosody and timbre across languages. This paper proposes VoiceGuard, a hybrid deep learning framework that combines complementary spectral and temporal rep-resentations for deepfake voice detection. A Convolutional Neural Network (CNN) branch learns frequency-domain artifacts from spectrograms, while a CNN-GRU branch models temporal inconsistencies from acoustic descriptors. An attention-based fusion mechanism adaptively weights branch outputs to improve discriminative power. The framework is evaluated on benchmark datasets and cross-lingual settings, and it improves performance compared to single-representation approaches while remaining compu-tationally practical for real-world deployment.

DOI: https://doi.org/10.5281/zenodo.19480877

 

Published by: vikaspatanker

Authors: Durgunala Ranjith, K.Hari Krishna, K.Rajender, R.Koti

Abstract: The project proposes an AI Startup Idea Validator that helps users evaluate startup ideas automatically using Artificial Intelligence and Large Language Models (LLMs). The system allows users to input their startup ideas through a web interface and analyzes them by considering factors such as market potential, competition, feasibility, and innovation. It uses AI-based processing to generate outputs including feasibility score, SWOT analysis, and improvement suggestions, providing users with clear insights into the strengths and weaknesses of their ideas. The system integrates external data sources and intelligent models to ensure accurate and data-driven decision-making. It is designed to be fast, cost-effective, and user- friendly, making it suitable for students, entrepreneurs, and startup incubators.

DOI: https://doi.org/10.5281/zenodo.19553722

 

Published by: vikaspatanker

Authors: Ambuj Kumar Misra

Abstract: The global agricultural sector faces unprecedented challenges from climate variability, resource depletion, and a rapidly growing population that demands consistent food security. This study presents the design, implementation, and evaluation of an Intelligent Agricultural Advisory System (IAAS) that leverages secondary crop datasets, multi-source meteorological records, and machine learning algorithms to deliver actionable, site-specific farming recommendations. Drawing on publicly available repositories including the USDA National Agricultural Statistics Service (NASS), NOAA Global Historical Climatology Network, and the FAO FAOSTAT database, our framework integrates data preprocessing pipelines, feature engineering modules, and ensemble predictive models comprising Random Forest classifiers and Long Short-Term Memory (LSTM) networks. Field validation across five Midwestern U.S. counties over a three-year period (2020-2023) demonstrated an average crop yield prediction accuracy of 91.4%, a 23.6% improvement in farmer decision-making efficiency, and a measurable reduction in water usage compared to conventional irrigation scheduling. The system's modular architecture supports deployment across a web dashboard and a mobile application accessible to smallholder and commercial farms alike. Our findings confirm that intelligent advisory systems built on secondary data are both technically feasible and economically significant, offering a scalable pathway toward precision agriculture for diverse agro-climatic regions.

DOI: https://doi.org/10.5281/zenodo.19479446

 

Published by: vikaspatanker

Authors: Ganesh Racha

Abstract: Financial technology is advancing rapidly, especially now since standard credit scoring methods are becoming obsolete. With scoring methods being archaic and out of touch, countless valuable behavioral data are not captured. In this study, the author discusses how possible behavioral data can be found in financial and transaction data using an AI-powered financial insight engine. It aims to change the predict and prescriptive analytics to enhance the better credit decision processes, beyond the usual finance means. Rather than referring to historical financial data and comparing it, behavioral data that is not ordinary are looked into particularly in expenditure. The result is a changing credit score that is indicative of the dynamic character of credit management. The use of advanced machine learning methods like Random Forest, Neural Networks and Gradient Boosting are remarkable in evaluating the above standard behavioral data and relationships, which are usually deemed to be irrelevant. The experimental results show that these models compared with other traditional methods like Logistic Regression are more accurate, precise and has better recall and score f1. In addition, the analysis of spending behavior has been integrated to introduce common financial user behavioral patterns and improve risk assessment and measurement of financial stability. An improved system demo is integrated that use cases widely for companies and banks. To similar how the companies were formed with tech such as Netflix, Samsung, Google and Uber changing the algorithm of credit check by enhancing AI algorithms along with blockchain records based validation, are used to analyse paticipants in this open eco-system.

DOI: https://doi.org/10.5281/zenodo.19479408

Published by: vikaspatanker