Enhanced Intrusion Detection System using Pearson Correlation based Long Short-Term Memory

Abstract

The Intrusion Detection System (IDS) is one of the common deep learning (DL) use cases which is used to finding and identifying outliers to prevent adversarial attacks, fraud and network intrusions. These systems play an important role in securing computer networks. The improvement of effective performance and security is the most needed in the IDS system. This paper aims to implement an IDS utilizing DL method. The DL model is the most encouraging technique and is widely used to detect intrusions. This paper implemented a Pearson's Correlation based Long Short-Term Memory (LSTM) model for intrusion detection system. The proposed Pearson's Correlation is evaluated on NSL-KDD dataset, and Min-Max normalization is a pre-processing method employed for normalize the data. Filter method of Pearson Correlation is utilized in the implemented for selecting the features. The LSTM is utilized to enhance system capability to identify and classify intrusion accurately and effectively. The implemented Pearson's Correlation based LSTM achieves better accuracy using NSL-KDD dataset with the value of 99.50% in term of accuracy, respectively.