Thettu Mokshagna Theja, M.Gowthami

Reinforcement Learning; Q-Learning; Warehouse Automation; Path Optimization; Grid-Based Navigation; Multi-Q-Table Architecture.

The rapid growth of warehouse automation has increased the demand for intelligent robotic systems capable of efficient and safe navigation in dynamic environments. Traditional path-planning algorithms perform well in static conditions but are less effective in environments with moving obstacles and multi-step tasks. To address this challenge, this paper presents RoboPath AI, a reinforcement learning-based warehouse robot path optimization system using tabular Q-learning. The proposed system models the warehouse as a 12×12 grid environment where the robot learns to navigate, collect multiple target items, avoid obstacles and human workers, and return to its starting position. A structured reward function is designed to encourage efficient movement and safe navigation. To improve learning performance, a multi-Q-table architecture is implemented, enabling the agent to handle sequential subtasks effectively. An epsilon-greedy strategy is used to balance exploration and exploitation during training. The system also incorporates real-time visualization, policy representation, and reward analysis to monitor learning progress. Additionally, Optuna-based hyperparameter tuning is applied to optimize performance. Experimental results demonstrate that the agent successfully learns optimal navigation strategies over time, showing significant improvement in efficiency and convergence.

[1] R. S. Sutton and A. G. Barto, Reinforcement Learning: An Introduction. MIT Press, 2018.
[2] C. J. C. H. Watkins and P. Dayan, “Q-Learning,” Machine Learning, vol. 8, no. 3–4, pp. 279–292, 1992.
[3] S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach. Pearson, 2021.
[4] V. Mnih et al., “Human-Level Control through Deep Reinforcement Learning,” Nature, vol. 518, pp. 529–533, 2015.
[5] D. Silver et al., “Mastering the Game of Go with Deep Neural Networks and Tree Search,” Nature, 2016.
[6] R. Bellman, Dynamic Programming. Princeton University Press, 1957.
[7] L. Busoniu et al., “Multi-Agent Reinforcement Learning: An Overview,” Innovations in Multi-Agent Systems, 2010.
[8] T. Schaul et al., “Prioritized Experience Replay,” ICLR, 2016.
[9] J. Schulman et al., “Proximal Policy Optimization Algorithms,” arXiv preprint, 2017.
[10] F. Hutter et al., “Sequential Model-Based Optimization for General Algorithm Configuration,” LION, 2011.
[11] T. Akiba et al., “Optuna: A Next-Generation Hyperparameter Optimization Framework,” KDD, 2019.
[12] M. L. Puterman, Markov Decision Processes. Wiley, 2014.
[13] S. Thrun et al., Probabilistic Robotics. MIT Press, 2005.
[14] P. Abbeel and A. Ng, “Apprenticeship Learning via Inverse Reinforcement Learning,” ICML, 2004.
[15] H. Kober et al., “Reinforcement Learning in Robotics: A Survey,” IJRR, 2013.

P. Hema, B. Shireesha

Smart Job Portal; ATS Score; Skill Matching; Artificial Intelligence; Resume Builder; Django; Recruitment Automation.

The rapid expansion of digital recruitment platforms has intensified the challenges associated with efficient candidate-job matching, manual resume screening, and lack of personalized career guidance. This paper presents a Smart Job Portal with AI-Powered Skill Matching and Career Assistance, a full-stack web-based system designed to streamline recruitment processes and enhance candidate experience through intelligent automation. The system is developed using the Django framework with Python as the core programming language and SQLite as the backend database. It incorporates an advanced Applicant Tracking System (ATS) that computes match scores by comparing candidate skills from both profile data and resume text against job requirements. The system further integrates AI-powered resume advice and chatbot assistance using large language model APIs such as OpenAI and Google Gemini. The platform supports multi-role access for candidates and HR professionals, providing dedicated dashboards and workflows. Candidates can build resumes, apply for jobs, track applications, and receive personalized suggestions, while recruiters can post jobs, evaluate applicants using ATS scores, and manage interview scheduling within a unified system. Experimental evaluation demonstrates improved recruitment efficiency, reduced manual effort, and enhanced candidate engagement. The proposed system offers a scalable and intelligent alternative to traditional job portals by combining automation, artificial intelligence, and full-stack development principles.

[1] Django Software Foundation, Django Documentation. [Online]. Available: https://docs.djangoproject.com/.
[2] OpenAI, OpenAI API Documentation. [Online]. Available: https://platform.openai.com/docs/.
[3] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding,” in Proc. 2019 Conf. North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT), Minneapolis, MN, USA, Jun. 2019, pp. 4171–4186.
[4] T. M. Mitchell, Machine Learning. New York, NY, USA: McGraw-Hill, 1997.
[5] Scikit-learn Developers, Scikit-learn Documentation. [Online]. Available: https://scikit-learn.org/stable/documentation.html..
[6] LinkedIn Corporation, LinkedIn Developer Documentation. [Online]. Available: https://learn.microsoft.com/linkedin/.
[7] Indeed, Indeed Hiring Lab: Labor Market Research and Hiring Trends. [Online]. Available: https://www.hiringlab.org/.
[8] Info Edge (India) Ltd., Naukri JobSpeak Report. [Online]. Available: https://www.naukri.com/jobsspeak.
[9] Bootstrap Team, Bootstrap Documentation. [Online]. Available: https://getbootstrap.com/docs/.
[10] SQLite Consortium, SQLite Documentation. [Online]. Available: https://www.sqlite.org/docs.html.
[11] Python Software Foundation, Python Documentation. [Online]. Available: https://docs.python.org/3/.
[12] Google, TensorFlow Documentation. [Online]. Available: https://www.tensorflow.org/api_docs.

Tholla Sunil, M. Gowthami

Event Management System; Django Framework; Web Application; Bootstrap; SQLite; Role-Based Access Control.

The increasing complexity of organizing events in academic, corporate, and community environments necessitates efficient and automated management systems. Traditional event coordination methods, which rely heavily on manual processes such as physical registrations, spreadsheets, and email communication, are prone to inefficiencies, errors, and scalability limitations. This paper presents a comprehensive Event Management System developed as a full-stack web application using the Django framework with SQLite as the backend database. The system provides a unified digital platform that supports two primary user roles: event organizers and participants. Organizers can create, manage, and schedule events, monitor participant registrations, track attendance, and send automated email notifications. Participants can browse available events, register for events, view schedules, and monitor their participation through a personalized dashboard. The system architecture follows the Model-View-Template (MVT) pattern, ensuring modularity, scalability, and maintainability. The application interface is designed using Bootstrap 5.3, providing a responsive and user-friendly experience across devices. The proposed solution effectively automates event coordination processes, reduces administrative overhead, enhances participant engagement, and provides a scalable platform for managing events in modern digital environments.

[1] D. Merkel, “Django: A High-Level Python Web Framework,” Linux Journal, vol. 2008, no. 168, pp. 1–10, 2008.
[2] Django Software Foundation, “Django Documentation,” 2024.
[3] M. Otto and J. Thornton, “Bootstrap Framework Documentation,” 2024.
[4] R. Elmasri and S. B. Navathe, Fundamentals of Database Systems, 7th ed. Pearson, 2016.
[5] SQLite Consortium, “SQLite Database Engine Documentation,” 2024.
[6] I. Sommerville, Software Engineering, 10th ed. Pearson, 2015.
[7] E. Gamma, R. Helm, R. Johnson, and J. Vlissides, Design Patterns. Addison-Wesley, 1994.
[8] W3C, “HTML5 Specification,” 2024.
[9] Mozilla Developer Network, “JavaScript Documentation,” 2024.
[10] S. Freeman and N. Pryce, Growing Object-Oriented Software. Addison-Wesley, 2009.
[11] P. Barry, Head First Python, 2nd ed. O’Reilly Media, 2016.
[12] T. Berners-Lee, “Information Management: A Proposal,” CERN, 1989.
[13] A. Silberschatz, H. Korth, and S. Sudarshan, Database System Concepts, 6th ed. McGraw-Hill, 2011.
[14] G. Booch, J. Rumbaugh, and I. Jacobson, The Unified Modeling Language User Guide. Addison-Wesley, 2005.
[15] J. Nielsen, Usability Engineering Principles. Academic Press, 1993.

Sankarapu Reddy Mahesh, S. Manjunath Reddy

Waste Classification; Deep Learning; MobileNetV2; Computer Vision; Explainable AI; Grad-CAM.

Waste management has emerged as a critical environmental concern due to rapid urbanization and increasing consumption patterns. Traditional waste segregation methods rely heavily on manual processes, which are inefficient, error-prone, and hazardous to human health. To address these challenges, this paper proposes RecycleVision AI, an intelligent waste classification system based on deep learning and computer vision techniques. The system utilizes the MobileNetV2 architecture with transfer learning to classify waste images into six categories: plastic, paper, glass, metal, cardboard, and general trash. A key contribution of this work is the integration of Explainable Artificial Intelligence using Grad-CAM, which provides visual insights into the model’s decision-making process by highlighting important image regions. The system is implemented using TensorFlow and deployed via a Streamlit-based web interface, enabling users to upload images and receive real-time classification results along with confidence scores and visual explanations. The proposed solution demonstrates high accuracy, computational efficiency, and usability, making it suitable for real-world waste management applications. It contributes to sustainable development by improving waste segregation efficiency and promoting responsible recycling practices.

[1] M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L. C. Chen, “MobileNetV2: Inverted Residuals and Linear Bottlenecks,” Proc. IEEE CVPR, 2018.
[2] R. R. Selvaraju et al., “Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization,” IEEE ICCV, 2017.
[3] M. Abadi et al., “TensorFlow: A System for Large-Scale Machine Learning,” OSDI, 2016.
[4] F. Chollet, “Keras: The Python Deep Learning API,” 2015.
[5] A. G. Howard et al., “MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications,” 2017.
[6] Y. LeCun, Y. Bengio, and G. Hinton, “Deep Learning,” Nature, 2015.
[7] O. Russakovsky et al., “ImageNet Large Scale Visual Recognition Challenge,” IJCV, 2015.
[8] K. Simonyan and A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” 2014.
[9] G. Bradski, “The OpenCV Library,” Dr. Dobb’s Journal, 2000.
[10] World Bank Group, “What a Waste 2.0: A Global Snapshot of Solid Waste Management,” 2018.

Pasupuleti Sai Deepthi, Mr. S. Manjunath Reddy

Employee Attrition Prediction; HR Analytics; Machine Learning (Random Forest); Workforce Performance Analysis; Predictive Modeling.

HR Insight Analytics is an end-to-end platform designed to predict employee performance and attrition risk, addressing the high costs and disruptions caused by unexpected workforce turnover. Attrition is influenced by multiple factors, including compensation dissatisfaction, limited career growth, poor work-life balance, workload stress, and weak managerial relationships. The platform leverages a synthetic dataset of 15,000 employee records with 34 attributes covering demographics, job roles, compensation, engagement levels, and attendance patterns. The system operates through four stages: a Python-based engine generates realistic workforce data; a Random Forest classification model with advanced feature engineering predicts attrition with high accuracy using cross-validation; data is structured in a normalized MySQL database for efficient storage and querying; and interactive dashboards present insights on attrition trends, performance distribution, and risk factors. The platform delivers actionable insights by assigning attrition probabilities, risk categories, and retention priorities to employees, along with tailored intervention strategies, enabling organizations to proactively manage workforce challenges and improve retention outcomes.

[1] L. Breiman, “Random Forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, 2001.
[2] F. Pedregosa et al., “Scikit-learn: Machine Learning in Python,” Journal of Machine Learning Research, vol. 12, pp. 2825–2830, 2011.
[3] W. McKinney, “Data Structures for Statistical Computing in Python,” Proc. 9th Python in Science Conf., 2010, pp. 51–56.
[4] S. M. Lundberg and S. I. Lee, “A Unified Approach to Interpreting Model Predictions,” Advances in Neural Information Processing Systems, vol. 30, pp. 4765–4774, 2017.
[5] N. V. Chawla et al., “SMOTE: Synthetic Minority Over-sampling Technique,” Journal of Artificial Intelligence Research, vol. 16, pp. 321–357, 2002.
[6] P. W. Hom et al., “Reviewing Employee Turnover: Focusing on Proximal Withdrawal States,” Psychological Bulletin, vol. 138, no. 5, pp. 831–858, 2012.
[7] T. R. Mitchell et al., “Why People Stay: Using Job Embeddedness to Predict Voluntary Turnover,” Academy of Management Journal, vol. 44, no. 6, pp. 1102–1121, 2001.
[8] J. L. Cotton and J. M. Tuttle, “Employee Turnover: A Meta-Analysis and Review,” Academy of Management Review, vol. 11, no. 1, pp. 55–70, 1986.
[9] Y. Zhao et al., “Enterprise Human Resources Analytics: A Literature Review,” Journal of Organizational Behavior, vol. 39, no. 8, pp. 1048–1060, 2018.
[10] Society for Human Resource Management, “SHRM Benchmarking Report: Human Capital,” Alexandria, VA, USA, 2022.
[11] C. R. Harris et al., “Array Programming with NumPy,” Nature, vol. 585, pp. 357–362, 2020.
[12] J. D. Hunter, “Matplotlib: A 2D Graphics Environment,” Computing in Science and Engineering, vol. 9, no. 3, pp. 90–95, 2007.
[13] G. van Rossum and F. L. Drake, Python 3 Reference Manual. Scotts Valley, CA, USA: CreateSpace, 2009.
[14] M. Bohanec et al., “Decision Making with Machine Learning and Rule-Based Classifiers,” Expert Systems with Applications, vol. 71, pp. 242–257, 2017.
[15] MySQL AB, “MySQL 8.0 Reference Manual,” Oracle Corporation, 2023.
[16] Microsoft Corporation, “Power BI Desktop Documentation,” 2024.
[17] W. H. Mobley, “Intermediate Linkages in the Relationship between Job Satisfaction and Employee Turnover,” Journal of Applied Psychology, vol. 62, no. 2, pp. 237–240, 1977.
[18] J. L. Price, The Study of Turnover. Ames, IA, USA: Iowa State University Press, 1977.
[19] R. Jain and A. Nayyar, “Predicting Employee Attrition Using XGBoost Machine Learning Approach,” Proc. Int. Conf. System Modeling and Advancement in Research Trends, 2018, pp. 113–120.
[20] D. Alao and A. B. Adeyemo, “Analyzing Employee Attrition Using Decision Tree Algorithms,” Computing, Information Systems, Development Informatics and Allied Research Journal, vol. 4, no. 1, pp. 17–28, 2013.

Sahni Sumith Kumar, S. Manjunath Reddy

Smart Agriculture; Django Framework; Crop Advisory; Web Application; Keyword-Based AI; Farmer Support System.

Agriculture plays a vital role in the economic development of countries like India, yet farmers often face challenges due to limited access to timely and accurate agricultural advisory services. Traditional knowledge-sharing methods are constrained by geographic, temporal, and resource limitations, leading to inefficiencies in farming practices. This paper presents a Smart Agriculture Advisory System, a web-based platform designed to bridge the gap between agricultural expertise and farmers through a centralized digital solution. The system is developed using a full-stack architecture with a frontend built using HTML, CSS, Bootstrap, and JavaScript, ensuring a responsive and user-friendly interface. The backend is implemented using Python and the Django framework, which manages application logic, user authentication, and database operations. SQLite3 is used as the database for storing farmer profiles, crop information, and query records. The platform consists of two main components: a Farmer Portal and an Admin Portal. Farmers can submit queries related to crop cultivation, fertilizers, pest management, and irrigation, while administrators manage crop data and advisory content. A keyword-based intelligent query processing engine generates real-time responses to farmer queries. The system improves accessibility to agricultural knowledge, enhances decision-making efficiency, and demonstrates the potential of web technologies in empowering rural communities.

[1] Django Software Foundation, “Django Documentation,” Version 5.x, 2024.
[2] Python Software Foundation, “Python Documentation,” 2024.
[3] Bootstrap Team, “Bootstrap 5 Documentation,” 2024.
[4] Mozilla Developer Network, “JavaScript Guide,” 2024.
[5] SQLite Consortium, “SQLite Documentation,” 2024.
[6] R. Elmasri and S. Navathe, Fundamentals of Database Systems, 7th ed. Pearson, 2016.
[7] I. Sommerville, Software Engineering, 10th ed. Pearson, 2015.
[8] S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 3rd ed. Pearson, 2016.
[9] J. Han, M. Kamber, and J. Pei, Data Mining: Concepts and Techniques, 3rd ed. Morgan Kaufmann, 2011.
[10] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. MIT Press, 2016.
[11] FAO, “Digital Agriculture Report,” 2022.
[12] Government of India, “Agricultural Extension Services Report,” 2023.
[13] OpenWeatherMap, “Weather API Documentation,” 2024.
[14] TensorFlow Developers, “TensorFlow Documentation,” 2024.
[15] Kisan Suvidha Portal, “Agricultural Information Services,” Government of India, 2023.

P. Soma Sekhar , B. Shireeshai, Dr. S. Usharani

Library Management System; Django; Automation; Borrowing Workflow; Penalty Calculation; AI Recommendation; Web Application.

The increasing demand for efficient information management in educational institutions has highlighted the limitations of traditional library systems, which rely heavily on manual processes for cataloguing, borrowing, and record maintenance. This paper presents a Smart Library Management System, a web-based solution designed to automate and optimize library operations while enhancing accessibility and user experience. The system is developed using the Django framework with Python as the backend language and SQLite for data persistence. The frontend interface is built using HTML5, CSS3, JavaScript, and Bootstrap, ensuring a responsive and user-friendly design. The system supports dual-role functionality, enabling library members to browse books, request borrowing, access digital resources, and receive notifications, while librarians manage inventory, approve requests, track loans, and handle overdue penalties. A key feature of the system is the automated borrowing workflow, which includes real-time book availability tracking, request approval mechanisms, and penalty computation for overdue returns. Additionally, an optional AI-powered recommendation module utilizing the Google Gemini API enhances user engagement by suggesting relevant books based on borrowing history. Experimental evaluation demonstrates improved operational efficiency, reduced administrative workload, and enhanced user satisfaction compared to traditional systems. The proposed system offers a scalable, cost-effective, and deployable solution for modern educational libraries.

[1] Django Software Foundation, Django Documentation. [Online]. Available: https://docs.djangoproject.com/. Accessed: Jul. 7, 2026.
[2] SQLite Consortium, SQLite Documentation. [Online]. Available: https://www.sqlite.org/docs.html.
[3] Google, Gemini API Documentation. [Online]. Available: https://ai.google.dev/gemini-api/docs.
[4] T. M. Mitchell, Machine Learning. New York, NY, USA: McGraw-Hill, 1997.
[5] Bootstrap Team, Bootstrap Documentation. [Online]. Available: https://getbootstrap.com/docs/.
[6] Python Software Foundation, Python Documentation. [Online]. Available: https://docs.python.org/3/.
[7] PostgreSQL Global Development Group, PostgreSQL Documentation. [Online]. Available: https://www. postgresql. org/docs/.
[8] Amazon Web Services, AWS Documentation. [Online]. Available: https://docs.aws.amazon.com/.
[9] Docker Inc., Docker Documentation. [Online]. Available: https://docs.docker.com/.
[10] The Kubernetes Authors, Kubernetes Documentation. [Online]. Available: https://kubernetes.io/docs/.
[11] K. Finkenzeller, RFID Handbook: Fundamentals and Applications in Contactless Smart Cards, Radio Frequency Identification and Near-Field Communication, 3rd ed. Hoboken, NJ, USA: Wiley, 2010.
[12] M. K. Buckland, “Barcode technology in libraries,” Library Hi Tech, vol. 8, no. 3, pp. 7–15, 1990.
[13] A. Breeding, “Library management systems: Current trends and future directions,” Computers in Libraries, vol. 42, no. 5, pp. 12–18, 2022.
[14] X. Liu, Y. Wang, and J. Zhang, “Digital library systems: Architecture, technologies and challenges,” Journal of Information Science, vol. 47, no. 6, pp. 755–770, Dec. 2021.
[15] IEEE Computer Society, IEEE Standard for Software and System Test Documentation, IEEE Std 829-2008, Jul. 2008.

Kommu Tharun, V.Vijayalakshmi, Dr.S.Usharani

Predictive Maintenance; Anomaly Detection; Autoencoder; Isolation Forest; Industrial IoT.

The rapid advancement of industrial automation and the increasing deployment of sensor-based monitoring systems have enabled data-driven approaches for intelligent equipment maintenance. This paper presents SmartSense AI, an industrial predictive anomaly detection framework designed to identify abnormal patterns in machine sensor data using advanced machine learning techniques. Traditional maintenance strategies, including reactive and preventive maintenance, often fail to detect early-stage equipment degradation, leading to unexpected failures and increased operational costs. The proposed system utilizes unsupervised learning methods, specifically an autoencoder-based deep learning model and the Isolation Forest algorithm, to analyze multivariate time-series sensor data. The autoencoder learns the normal behavior of machine operations and identifies anomalies based on reconstruction error, while the Isolation Forest detects statistically isolated data points. A dual-model approach enhances detection reliability and reduces false positives. Additionally, the system incorporates a severity scoring mechanism to quantify anomaly intensity and a root cause analysis module to identify contributing sensor variables. An interactive Streamlit-based dashboard enables real-time visualization of anomaly detection results, severity levels, and sensor insights. The system is implemented using Python, TensorFlow, and Scikit-learn, ensuring scalability and efficiency. Experimental evaluation demonstrates that SmartSense AI effectively detects early anomalies, reducing unplanned downtime and maintenance costs. The modular architecture supports future enhancements such as IoT integration, cloud deployment, and predictive failure estimation.

[1] Chandola, V., Banerjee, A., Kumar, V., “Anomaly Detection: A Survey,” ACM Computing Surveys, Vol. 41, No. 3, 2009, pp. 1–58.
[2] Liu, F. T., Ting, K. M., Zhou, Z. H., “Isolation Forest,” IEEE International Conference on Data Mining, 2008, pp. 413–422.
[3] Sakurada, M., Yairi, T., “Anomaly Detection Using Autoencoders with Nonlinear Dimensionality Reduction,” MLSDA, 2014.
[4] Pang, G., Shen, C., Cao, L., Hengel, A., “Deep Learning for Anomaly Detection: A Review,” ACM Computing Surveys, 2021.
[5] Chalapathy, R., Chawla, S., “Deep Learning for Anomaly Detection: A Survey,” arXiv preprint, 2019.
[6] Breunig, M. M., Kriegel, H. P., Ng, R. T., Sander, J., “LOF: Identifying Density-Based Local Outliers,” SIGMOD, 2000.
[7] Schölkopf, B., Platt, J., Shawe-Taylor, J., Smola, A., Williamson, R., “Estimating the Support of a High-Dimensional Distribution,” Neural Computation, 2001.
[8] An, J., Cho, S., “Variational Autoencoder Based Anomaly Detection,” Special Lecture on IE, 2015.
[9] Goodfellow, I., et al., “Generative Adversarial Nets,” NIPS, 2014.
[10] Carvalho, T. P., et al., “A Systematic Literature Review of Machine Learning Methods Applied to Predictive Maintenance,” Computers & Industrial Engineering, 2019.
[11] Khan, S., Yairi, T., “A Review on the Application of Deep Learning in System Health Management,” Mechanical Systems and Signal Processing, 2018.
[12] Ahmad, S., Lavin, A., Purdy, S., Agha, Z., “Unsupervised Real-Time Anomaly Detection for Streaming Data,” Neurocomputing, 2017.

Boochi Adikeshavulu, Dr. S. Usharani

Graph Neural Networks; Fraud Detection; Fraud Rings; Transaction Graph; Graph Attention Networks; GCN; GAT; GraphSAGE; Financial Security.

Financial fraud continues to evolve with increasing sophistication, particularly through coordinated fraud rings that exploit distributed transaction patterns to evade detection. Traditional rule-based systems and tabular machine learning models are limited in capturing relational dependencies across entities such as accounts, devices, and identities. This paper presents FraudNet AI, a graph-based deep learning framework that models financial transactions as interconnected nodes within a transaction graph to identify complex fraud structures. The proposed system utilizes the IEEE-CIS Fraud Detection dataset and constructs a heterogeneous graph by linking transactions through shared attributes including device identifiers, IP addresses, billing information, and email domains.A hybrid Graph Neural Network (GNN) architecture combining Graph Convolutional Networks (GCN), Graph Attention Networks (GAT), and GraphSAGE is employed to learn rich node embeddings that encode both intrinsic transaction features and neighborhood relationships. The framework includes preprocessing, graph construction, model training with class imbalance handling, and deployment via a Flask-based API and interactive dashboard. Experimental results demonstrate improved performance over baseline models in terms of AUC-ROC, precision, and recall.

[1] T. N. Kipf and M. Welling, “Semi-Supervised Classification with Graph Convolutional Networks,” International Conference on Learning Representations (ICLR), 2017.
[2] P. Velickovic et al., “Graph Attention Networks,” International Conference on Learning Representations (ICLR), 2018.
[3] W. Hamilton et al., “Inductive Representation Learning on Large Graphs,” Advances in Neural Information Processing Systems (NeurIPS), 2017.
[4] T. Chen and C. Guestrin, “XGBoost: A Scalable Tree Boosting System,” ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), 2016.
[5] L. Breiman, “Random Forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, 2001.
[6] M. Fey and J. E. Lenssen, “Fast Graph Representation Learning with PyTorch Geometric,” ICLR Workshop on Representation Learning on Graphs and Manifolds, 2019.
[7] IEEE-CIS, “Fraud Detection Dataset,” Kaggle Competition Dataset, 2019.
[8] A. Paszke et al., “PyTorch: An Imperative Style, High-Performance Deep Learning Library,” Advances in Neural Information Processing Systems (NeurIPS), vol. 32, 2019.
[9] F. Pedregosa et al., “Scikit-learn: Machine Learning in Python,” Journal of Machine Learning Research (JMLR), vol. 12, pp. 2825–2830, 2011.
[10]Z. Ying et al., “GNNExplainer: Generating Explanations for Graph Neural Networks,” Advances in Neural Information Processing Systems (NeurIPS), 2019.

Bonasi Poojitha, Dr.S.Usharani

Ocular Disease Detection; Deep Learning; EfficientNet, Fundus Imaging; Grad-CAM, Diabetic Retinopathy; Explainable AI.

Vision impairment and blindness caused by ocular diseases remain a major global health challenge, affecting over 2.2 billion people worldwide. Many conditions such as diabetic retinopathy, glaucoma, cataract, age-related macular degeneration, hypertension-related retinopathy, and myopia are progressive but preventable with early diagnosis. However, limited access to ophthalmologists, high consultation costs, and time-intensive manual diagnosis hinder timely detection, especially in developing regions. RetinaVision AI addresses these challenges through a deep learning-based automated system for retinal disease detection using fundus images. The system is built on the EfficientNet-B3 architecture, known for its high accuracy and computational efficiency. It employs two models: a multi-disease classifier that detects eight retinal conditions, and a specialized model for diabetic retinopathy severity grading across four levels.To enhance interpretability, the system integrates Grad-CAM, which generates heatmaps highlighting important regions in the retinal image, ensuring clinically relevant decision-making. RetinaVision AI is deployed as a user-friendly web application using Streamlit, enabling easy access through a browser. Overall, the system demonstrates the effective use of deep learning, transfer learning, and explainable AI for scalable and reliable ophthalmological diagnostics.

[1] M. Tan and Q. V. Le, “EfficientNet: Rethinking model scaling for convolutional neural networks,” Proc. ICML, vol. 97, pp. 6105–6114, 2019.
[2] V. Gulshan et al., “Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs,” JAMA, vol. 316, no. 22, pp. 2402–2410, 2016.
[3] R. R. Selvaraju et al., “Grad-CAM: Visual explanations from deep networks via gradient-based localization,” Proc. ICCV, pp. 618–626, 2017.
[4] D. S. W. Ting et al., “Development and validation of a deep learning system for diabetic retinopathy and related eye diseases using retinal images from multiethnic populations with diabetes,” JAMA, vol. 318, no. 22, pp. 2211–2223, 2017.
[5] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” Proc. CVPR, pp. 770–778, 2016.
[6] K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” arXiv preprint arXiv:1409.1556, 2014.
[7] J. Deng et al., “ImageNet: A large-scale hierarchical image database,” Proc. CVPR, pp. 248–255, 2009.
[8] A. Paszke et al., “PyTorch: An imperative style, high-performance deep learning library,” Adv. Neural Inf. Process. Syst., vol. 32, pp. 8026–8037, 2019.
[9] Aneeshkumar A. S., “Relevance of Data Mining for Presumption of Anarchy,” International Journal of Linguistics and computational Applications (IJLCA), IRDP Group of Journals, vol.3, issue 4, pp. 72-74, 2016.
[10] G. Bradski, “The OpenCV library,” Dr. Dobb’s J. Softw. Tools, vol. 25, no. 11, pp. 120–125, 2000.
[11] S. van der Walt, S. C. Colbert, and G. Varoquaux, “The NumPy array: A structure for efficient numerical computation,” Comput. Sci. Eng., vol. 13, no. 2, pp. 22–30, 2011.
[12] World Health Organization, World Report on Vision. Geneva, Switzerland: WHO Press, 2023.
[13] A. Esteva et al., “Dermatologist-level classification of skin cancer with deep neural networks,” Nature, vol. 542, pp. 115–118, 2017.
[14] Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, pp. 436–444, 2015.
[15] Streamlit Inc., “Streamlit documentation,” 2024. [Online]. Available: https://docs.streamlit.io
[16] A. Dosovitskiy et al., “An image is worth 16×16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929, 2020.
[17] B. Zhou et al., “Learning deep features for discriminative localization,” Proc. CVPR, pp. 2921–2929, 2016.
[18] N. V. Chawla et al., “SMOTE: Synthetic minority over-sampling technique,” J. Artif. Intell. Res., vol. 16, pp. 321–357, 2002.
[19] Kaggle, “APTOS 2019 blindness detection dataset,” 2019. [Online]. Available: https://www.kaggle.com/c/aptos2019-blindness-detection
[20] T. Li et al., “Applications of deep learning in fundus images: A review,” Med. Image Anal., vol. 69, p. 101971, 2021.
[21] M. D. Abramoff et al., “Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices,” npj Digital Medicine, vol. 1, p. 39, 2018.