Network Intrusion Classifier Using Autoencoder with Recurrent Neural Network

In the last few decades, the neural network has been solving a variety of complex problems in engineering, science, finance, and market analysis. Among them, one of the important problems is a protection system against of threat of cyber-attacks. In this paper, we have proposed the deep learning approach where Sparse Autoencoder (SAE) and Recurrent Neural Network (RNN) are combined for the detection of network intrusion. We evaluate the proposed approach based on different performance metrics by applying it to the NSL-KDD dataset. Through the performance test, the proposed method achieved high accuracy in the intrusion detection.