The corporate system allows identifying possible threats associated with card transactions at ATMs, payment terminals, etc. A threat is considered uncharacteristic behavior for a user. For example, conducting a large number of transactions with a small amount of money that does not require entering a pin code, or transactions not previously conducted for this product group. When a threat is identified, the system transmits this data in real-time to the administrator.
What Was Done
To create this solution, we identified the factors that allow anomalies to be detected and integrated software and hardware sensors into the equipment system. Then we analyzed the initial anomalies and created a training environment to train the system. Next, we set up an anomaly alert module and an analytics module, which allows you to identify and recognize the cause of deviation.
Technologies and tools: Python, PyTorch, AutoEncoder, SVM.