Secure network intrusion detection system using NID-RNN based Deep Learning

Towards Secure network intrusion detection system (NIDS) is an important protection appliance in cyber security. Traditional methods that employ machine learning techniques necessitate the selection of features manually, which has obvious drawbacks. Nowadays, deep learning is one of trending technology that intrusion detection models are used to increase performance in securing networks and to detect the malicious attacks. Its quick learning and capability for sequential learning in its network intrusion detection (NID) form are encouraging factors for its use. The unpredictable weights of the input-hidden layer are an issue with NID that researchers have addressed. As a result, there is concern about both the convergence and the speed. It describes the NID-Recurrent neural network (RNN) approach for integrating NID with Long short term memory (LSTM). On the one hand, the technique is tested using different activation function scenarios for NID, and on the other, the number of iterations for BP. The UNSW-NB18 datasets are used to evaluate the performance of the proposed model based on both binary and multiclass classifications. The experimental results show that the accuracy of proposed RNN was increased by about 8%, compared with existing RNN algorithm in order to reach the optimal accuracy with a little number of iterations.