Deep Learning Approach for Detection of Fraudulent Credit Card Transactions

Instead of cash, people tend to use credit cards with the swift technological growth in the modern world. This unlocks the door for fraudulent individuals to utilize these cards in a wicked method. Every year, it costs billions of dollars in credit card transaction fraud to card issuers. There are no static patterns in fraud. Their behavior constantly changes. New technologies allow fraudsters to use the online medium and other techniques for implementing frauds. It is vital to learn the behavior patterns. The detection accuracy can be increased with a large dataset and complex features. This chapter addresses the problem of analyzing fraudulent credit card transactions. Explicitly, we propose a deep learning-based framework with various unsupervised learning algorithms. We perform the Hyperparameter optimization of these algorithms using sci-kit-learn machine learning frameworks and popular deep learning framework TensorFlow. In the end, we discussed the applied implementation of detecting the fraudulent transactions on the real-world dataset available on Kaggle.