Intrusion Detection System: Overview

TL;DR

This paper reviews four different network intrusion detection approaches—ANN, SOM, Fuzzy Logic, and SVM—highlighting their methodologies and exploring hybrid systems combining supervised and unsupervised learning techniques.

Contribution

It provides an overview of various IDS methods and discusses the potential for hybrid approaches combining supervised and unsupervised learning.

Findings

ANN is a traditional supervised IDS method

SOM and Fuzzy Logic are unsupervised approaches

Hybrid IDS approaches are promising for improved detection

Abstract

Network Intrusion Detection (NID) is the process of identifying network activity that can lead to the compromise of a security policy. In this paper, we will look at four intrusion detection approaches, which include ANN or Artificial Neural Network, SOM, Fuzzy Logic and SVM. ANN is one of the oldest systems that have been used for Intrusion Detection System (IDS), which presents supervised learning methods. However, in this research, we also came across SOM or Self Organizing Map, which is an ANN-based system, but applies unsupervised methods. Another approach is Fuzzy Logic (IDS-based), which also applies unsupervised learning methods. Lastly, we will look at the SVM system or Support Vector Machine for IDS. The goal of this paper is to draw an image for hybrid approaches using these supervised and unsupervised methods.