On Detecting Frauds in Retail Transactions for Enhanced Security with Deep Learning Model

Retail fraud poses substantial challenges to an organisation’s financial viability and image. This work describes a deep learning-based fraud detection and prevention system that uses models such as RNN, CNN, LSTM, and GAN. Our technique entails analysing transaction data, such as cash in, cash out, and balance variations, to increase the efficiency of fraud detection. The Convolutional Neural Network outperforms other models with 94.46% accuracy, 98% precision, 90% F1 score, and 94% recall. This system leverages these deep learning models to equip merchants with proactive tools for early fraud detection and prevention.