MUMBAI, India, July 13 -- Intellectual Property India has published a patent application (202621064237 A) filed by Symbiosis International Deemed University on May 21, 2026, for Hybrid Autoencoder And K-Means Clustering System For Real-Time Fraud Detection In E-Payment Networks.
Inventors include Ram Wadatkar; and Mrinal Bachute.
The application for the patent was published on July 10, 2026, under issue no. 28/2026.
Abstract: ABSTRACT Hybrid Autoencoder and K-Means Clustering System for Real-Time Fraud Detection in E-Payment Networks The present invention provides a hybrid artificial intelligence system (100) for real-time fraud detection in high-velocity electronic payment networks. The system comprises a Transaction Ingestion Gateway (101), a pre-processing pipeline including a SMOTE Module (102a) for class balancing and a PCA Module (102b) for dimensionality reduction, a Stage 1 Deep Autoencoder (103) generating a Reconstruction Error (104) and Latent Space Representation (105), a Stage 2 K-Means Clustering Engine (106) with PSO Engine (107) calculating a Spatial Distance (108), a Hybrid Scoring Logic Engine (109) combining structural and behavioural anomaly indicators to produce a Hybrid Anomaly Score, and an Authorization Gateway (110) issuing real-time Approve or Decline signals. The system operates in an unsupervised paradigm without labelled fraud datasets, detects zero-day attacks, supports decentralised edge-computing deployment with sub-50 millisecond inference latency, and achieves 92 percent precision and 0.94 AUC-ROC. [
Disclaimer: Curated by HT Syndication.
