A CNN-RF Hybrid Model for Intrusion Detection System: Analysis, Improvements, and Application

Abstract

With the rapid development of technologies and the growing need for the internet in most aspects of life, the need for cyber security is also growing. There are various types of cyber intrusions attacks making detecting and identifying these attacks not an easy task. Conventional intrusion detection systems (IDS) lack accuracy, which makes them unreliable and dependable. Recent applications of machine learning techniques have rapidly grown in the recent years, made it a powerful tool that can be utilized for detecting cyber intrusion attack accurately. This paper proposed an enhanced convolution neural networks (CNNs)-based machine learning model for intrusion detection. This model makes use of the characteristics of CNN layers for extracting useful features and the Random Forest model for Robust intrusion attack detection. The model utilizes the NSL-KDD dataset, and it outperforms other deep learning (DL) techniques in the multi-class classification tasks, it can identify intrusion attacks with high precision and reach 99.3% accuracy rate leading for increasing the efficiency of intrusion detection and open new avenues for research.