Smart Intrusion Detection Systems Using Machine Learning

Abstract

Engineering problems are generally solved by gathering domain specific knowledge and then developing algorithms to solve that particular problem. Machine learning is an Engineering discipline that helps us solve these problems by collecting a large datasets and then with the help of that data it is possible to create black box machine output of which can then be used to solve real world problems [1]. This paper cites the basis of machine learning concepts and the methods which can be applied to intrusion detection in the cyber security industry. Machine learning is used to generate predictive models that can be used to classify requests on to a server. Main goal is to build a Machine learning model that can classify a request on a server into a normal request or a malicious request based on the server’s behaviour and determine the type of attack. Proposing a new system that employs data science community such as normalization through the minmax algorithm, feature selection through the chi square method, planning to separate the data by the Gaussian kernel and then approaching to employ two machine learning models.

Keywords: Chi square, Cyber security, Gaussian kernel, Intrusion detection system, Machine learning, Pattern recognition, SVM, Training data.

View PDF