Autonomous credit card fraud detection using machine learning approach

Credit card fraud has risen in vulnerable effects in recent years as more people use credit cards to pay for products. This is owing to advancements in technology and growths in internet transactions, both of which have resulted in massive financial losses due to fraud. To reduce such losses, an effective fraud detection system must be designed and implemented. Machine learning approaches used to detect credit card fraud automatically and do not take into account deception process or behavioral problem, which might lead to alerts. The goal of this study is to figure out how to spot credit card fraud. To detect the occurrence of fraud, a Long Short-Term MemoryRecurrent Neural Network (LSTM-RNN) is proposed. In addition, an attention mechanism has been included to increase performance even more. In instances like fraud detection, where the information sequence is made up of vectors with complicated interrelated properties, models with this structure have proven to be particularly efficient. LSTM-RNN is compared to other classifiers such as Naive Bayes, Support Vector Machine (SVM), and Artificial Neural Network (ANN). Experiments reveal that our proposed model produces powerful results and has a high level of accuracy.