Intellectual Property India Publishes Patent Application for A Method For Iterative Explainable Quantum Machine Learning-Based Adaptive Fraud Detection In Financial Transaction Systems

MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202641084476 A) filed by Sr University; and . Email M. Sheshikalasru. Edu. In Mobile Telangana India on July 09, 2026, for A Method For Iterative Explainable Quantum Machine Learning-Based Adaptive Fraud Detection In Financial Transaction Systems.

Inventors include Mr. Mudimela Madhusudhan; Dr. Pramoda Patro; and Dr. Vishwanath Bijalwan.

The application for the patent was published on July 17, 2026, under issue no. 29/2026.

Abstract: Abstract The present invention relates to a method for adaptive fraud detection in financial transaction systems using iterative explainable quantum machine learning techniques. The method comprises receiving and preprocessing financial transaction data to generate quantum-compatible transaction representations. The method further comprises encoding the transaction data into a quantum latent space using a Hybrid Quantum-Loss Adaptive Variational Autoencoder (HQLA-VAE) for anomaly detection based on reconstruction loss and quantum fidelity evaluation. A Quantum Feature Relevance Graph Attribution (Q-FRGA) method performs graph-based fraud analysis using quantum graph neural networks to generate node-level and graph-level relevance explanations associated with transaction entities. The method further comprises adaptively refining fraud detection policies using a Quantum Environment-Adaptive Policy Refinement (Q-EAPR) process based on reinforcement learning and quantum Bellman updates for evolving fraud behavior analysis. Counterfactual fraud explanations are generated using a Cross-Modal Quantum Counterfactual Explainer (CQCE) configured to produce interpretable transaction-level explanation outputs. Outputs from the quantum anomaly detection, graph attribution, adaptive policy refinement, and counterfactual explanation processes are aggregated using a Quantum Residual Explainability-Integrated Ensemble Framework (Q-REIEF) to generate final fraud prediction outputs with explanation fidelity and confidence scoring. The proposed method improves fraud detection accuracy, interpretability, temporal adaptability, anomaly sensitivity, and real-time fraud detection capability in digital financial transaction environments. Accompanied Drawing [FIG. 100]

Disclaimer: Curated by HT Syndication.