An Autoencoder and LSTM based Intrusion Detection approach against Denial of service attacks

Abstract

The advent of technology made the nations to grow in a rapid phase but most of the nation are not ready to defend the critical infrastructure cyber-attacks on government databases and many organizations. Several cyber-attacks, in recent days Denial of Service (DoS) attacks are popular because of their severe impact on the network and its resources. Anomaly detection has been a field of intense research over the years as it poses many challenging problems. Machine learning and Deep Learning techniques have proven to be useful in identifying the anomalous patterns with least number of false positives. In this paper we present an Intelligent IDS built using advanced artificial neural network algorithms such as Autoencoders and Long Shor-Term Memory (LSTM). The proposed model is a novel approach which eliminates the challenges with time recurrent neural network architecture such as the response time in backpropagation. The LSTM algorithm is derived from the deep learning, which has shown promising results to learn and detect novel attacks.