Abstract: Credit card fraud is the most serious form of risk to financial security and it has a far-reaching effect on consumers and users in general; this makes it necessary that modern methods for machine learning like neural networks should be employed in order to improve their capacity and accuracy for detecting frauds. Various popular neural network architectures for fraud detection, including Feedforward Neural Networks, Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory Networks (LSTM), Autoencoders, and Generative Adversarial Networks (GAN) are investigated in this study.