Lightweight Neural Network based UPI Fraud Detection for Transactional Security

This document introduces a lightweight fraud detection system powered by a neural network, specifically developed to safeguard real-time transactions within the Unified Payments Interface (UPI). As digital payment systems become increasingly popular, the threat of fraud has grown, highlighting the need for smart and effective detection solutions. The proposed system employs a neural network model that has been trained on transactional and behavioral characteristics to accurately pinpoint suspicious activities. Essential components of the system encompass data annotation, environment configuration, neural prediction, and real-time visualization using a web interface. It is designed to function in low-latency settings while providing efficient fraud detection and maintaining computational resource efficiency. The findings underscore its appropriateness for scalable, practical implementation in contemporary digital payment frameworks.