Intellectual Property India Publishes Patent Application for A System And A Method Of Privacy-Preserving Cross-Bank Financial Fraud Detection Using Explainable Federated Learning

MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641078105 A) filed by Dr. M. G. R. Educational & Research Institute on June 24, 2026, for A System And A Method Of Privacy-Preserving Cross-Bank Financial Fraud Detection Using Explainable Federated Learning.

Inventors include Anusha C; Jashwanth G; Dahlia Sam; A. Kameshwaran; and N. Kanya.

The application for the patent was published on July 03, 2026, under issue no. 27/2026.

Abstract: Disclosed is a system (100) for privacy-preserving cross-bank financial fraud detection. The system (100) includes a plurality of bank nodes (102), a central aggregator server (120), a security operations center dashboard (134), a processor (140), and a memory (142). Each bank node retains private customer transaction data in a local transaction data repository (104), generates fraud-relevant feature values through a local feature engineering module (106), and trains a local machine learning model (108). A model update extraction module (112) extracts model update information, a differential privacy module (114) injects calibrated statistical noise, and a secure multi-party computation module (116) cryptographically masks the update information. A hybrid privacy layer (118) protects the updates before aggregation by a federated averaging module (122). A fraud prediction module (126) generates fraud risk outputs, and an explainable artificial intelligence engine (128) with a SHAP-based explanation module (130) generates human-readable audit reasoning.

Disclaimer: Curated by HT Syndication.