Abstract: Credit card fraud poses a significant threat to financial institutions and consumers alike. The ability to accurately detect fraudulent transactions is crucial for mitigating financial losses and protecting customers. In this paper, we explore the application of autoencoder neural networks for anomaly detection in credit card transactions. Autoencoders, a type of unsupervised deep learning model, have shown promising results in capturing complex patterns and identifying anomalies in various domains. We propose an approach that leverages the power of autoencoders to unveil anomalies within credit card transaction data. Through extensive experimentation and evaluation, we demonstrate the effectiveness of our approach in detecting fraudulent activities with high precision and recall rates. Furthermore, we discuss the interpretability and scalability of the autoencoder-based anomaly detection system and highlight potential areas for future research and improvements.

| DOI: 10.17148/IARJSET.2023.10608