Privacy-preserving Multiple Payment Fraud Detection Using Lstm-based Federated Learning.

Abstract

This Paper Presents A Privacy-preserving Framework For Multiple Payment Fraud Detection Using Long Short-Term Memory (LSTM) And Federated Learning. The Model Captures Temporal Patterns In Transaction Data To Identify Fraudulent Activities Effectively. Federated Learning Enables Decentralized Training Across Multiple Clients Without Sharing Sensitive Data, Ensuring Privacy And Security. Experimental Results Show That The Proposed Approach Achieves Over 95% Accuracy With Improved Precision And Recall Compared To Traditional Methods. Additionally, It Reduces Data Leakage Risks While Maintaining Scalability And Efficiency. The Proposed System Is Suitable For Real-time Fraud Detection In Distributed Financial Environments.

Keywords