Toward Deep Learning based Intrusion Detection System: A Survey
Zhiqi Li, Weidong Fang, Chunsheng Zhu, Guannan Song, Wuxiong Zhang
Abstract
As a pivotal defense mechanism against cyber-attacks, the intrusion detection system (IDS) is widely recognized. The remarkable accuracy exhibited by IDS in detecting various types of intrusions, owing to the leverage of deep learning (DL), prompts a surge in research endeavors aimed at DL-based IDS design. To facilitate researchers' access to the latest breakthroughs, we delve into recent advancements in DL-based IDS proposed over the past years. These works are systematically categorized into two main application domains: computer networks and the Internet of Things (IoT), and their methodology, accuracy performance, advantages, and disadvantages undergo scrutiny in each work, fostering an insightful comparison. Subsequently, meticulous examination and deliberation are conducted on the shared traits and distinctive features across these works. Drawing from the collective insights gleaned from the reviewed literature, the current developmental landscape is synthesized, and prospective research directions for future works are delineated in the conclusion.