Comparison of Recurrent Neural Network Algorithms for Intrusion Detection Based on Predicting Packet Sequences

Recurrent neural networks (RNN) shows a remarkable result in sequence learning, particularly in architectures with gated unit structures such as long short-term memory (LSTM). In recent years, several permutations of LSTM architecture have been proposed mainly to overcome the computational complexity of LSTM. In this paper, we present the first study that will empirically investigate and evaluate LSTM architecture variants specifically on a intrusion detection dataset. The investigation is designed to identify the learning time required for each LSTM algorithm and to measure the intrusion prediction accuracy. The results show that each variant exhibit improvement at specific parameters, yet, with a large dataset and short time training, none outperformed the standard LSTM.