Deep Learning for Intrusion Detection

5Intrusion Detection Through Deep Learning: Emerging Trends and Challenges

Achyutananda Mishra

School of Law, Christ University, Bangalore, Karnataka, India

5.1 Introduction

Internet technology has brought the world closer than ever seamlessly connecting all parts of the world. Its increasing use has exposed the world to cyberattacks (malicious attacks)—manual or machine-generated—compromising the data and system/network security of individuals and institutions. Malicious cyberattacks are a serious security threat; malware attacks are dynamic in nature; they keep changing so does their methodology of attacking [1]. Cybercrime is on the rise, cybercriminals operate internationally with ease disrupting businesses and services, and malware is intentionally created, compromising the security of computer systems [2].

The intrusion detection system (IDS) has emerged as a network security tool to monitor computer networks or systems for malicious attacks and detect intrusion. It can be a network intrusion detection system (NIDS), a host intrusion detection system (HIDS), or a cloud intrusion detection system. IDSs commonly employ techniques like signature-based detection, anomaly-based detection, and stateful protocol analysis. It is a software that observes network traffic to detect intrusion, and if it detects any unusual activity or policy violation, then it immediately sends an alert or blocks it. It is built on a predictive model (a classifier) to distinguish between “bad connections” ...