Anomaly Based Intrusion Detection System using Machine Learning and Neural Networks

Abstract

As we know in today’s era, the internet is an essential pillar on which the world resides. With the advancement in technology, network speed, we also need higher processing power, which should be reliable and feasible to use, we become more dependent on the internet. With this dependency, there is a continuously growing concern about the security and privacy of data. We also do not want to compromise with a fast server and response time. With this increase we also show an increase in Intrusion happening around the world. Intrusions are getting more sophisticated, more frequent day-by-day making a simple intrusion detection method, and exclusive Intrusion detection methods are no longer an option. We require advanced techniques to detect intrusion, which results in bulkier Intrusion Detection Systems. Currently, we need a resource-efficient Intrusion Detection System with no compromise with security and have a good response time. We have used classical machine learning algorithms and deep learning techniques to build the mlodel for classification and then try to see which all models are having better accuracy than existing systems. The dataset that we are using is NSL-KDD. We also took this problem as a binary as well as multi-class classification problem.