Hybrid CNN- BILSTM-Attention Based Identification and Prevention System for Banking Transactions

Abstract

With the increase in digitalization and more pressure on cashless transactions being put forth by our government there is a major risk of the credit card fraud in budgetary exchanges. Hence, the credit card fraud detection is the utmost responsibility and a crucial challenge up here for researchers to lessen the incurring losses by the banks and customers. The Customary detection method by organizations would mainly rely on rules. But there are some limitations to such rule-based methods as each rule has a corresponding threshold and being absolute in nature, they are inefficient when used alone. Moreover, they cannot treat a large amount of data at the same time. Nowadays, researchers are laboring hard into techniques such as Artificial Intelligence, Machine Learning, Data Mining, etc. In this paper, I have used Deep Learning techniques like CNN, BILSTM with ATTENTION layer. Using these techniques, the maximum efficiency which could be attained is 95%. Analytical outcomes applied to the real-world transaction information given by a business bank shows that our proposed strategy achieve desirable accuracy than any other techniques.