Project: AI Eagle Eye Fraud System Federated Learning (Eagle Eye) Qorus-Infosys Finacle Banking Innovation Awards 2024

The model's core functionalities include providing early warning provisions through various communication channels such as physical branches, internet/mobile platforms, and ATMs. It is also equipped with real-time hotspot detection, which enables immediate notification to the police network to apprehend money mules. The model utilizes pattern recognition and advanced data analytics techniques to thoroughly analyze transactions and account-holder behaviors, assigning a risk score to customer accounts based on the likelihood of fraudulent activity.

When the risk score surpasses a predetermined threshold, typically set at >95% confidence, it generates an alert. After flagging, our staff reviews the account to decide whether to confirm its status. Once confirmed, it is restricted from receiving fund transfers, and any attempts to withdraw from ATMs trigger notifications to the police for further investigation. This swift intervention by our staff ensures immediate action is taken to safeguard our customers' finances.

During the interim period between alert escalation and confirmation, customers making fund transfers are reminded of the pervasive nature of fraudulent activities, encouraging greater vigilance. Subsequently, the outcomes are evaluated to assess the efficacy of the flagged status, and necessary modifications are made to the model, facilitating a continuous feedback loop. This iterative approach ensures the model's ongoing improvement and effectiveness in combating fraudulent activities.