A Scalable Hybrid Model for Detecting UPI Transaction Fraud in Real-Time

The Unified Payments Interface (UPI) has revolutionised digital commerce by making fund transfers quick and simple. However, an increase in scams affecting unaware consumers has also emerged from this convenience. This study proposes a machine learning-based detection approach that analyses the activity of transactions to identify problematic patterns in order to address the rising issue of electronic payment scam. Applying Hidden Markov Models (HMM), which are effective at simulating consecutive information and spotting variations from typical transaction trends, is the cornerstone of the methodology. The system can detect subtle behavioural shifts that can be indicators of dishonesty via learning from historical information on users, enabling timely detection and response. Unlike static, rule-driven methods, the proposed technique adapts to evolving deception strategies, improving detection precision and lowering false positives. The ultimate goal is to offer a robust, intelligent platform that strengthens UPI transaction safety, reduces financial risk, and boosts trust in digital payment systems.

Ҳали таржима қилинмаган