Optimizing LightGBM for Intrusion Detection Systems using GOA

Increasing popularity of IoT devices across the world is because of its applicability in wide range of areas and disciplines like healthcare, education, business, defense etc. The information is continuously being shared as they work 24x7, thus making vulnerable to several types of attacks. Intrusion Detection Systems are smart intelligent systems that detects the occurrence of any malicious activity. Due to the growing size of the internet, intruders are trying new ways to inject newer kinds of attacks and perform malicious activities. Hence, there is a need of robust and reliable IDSs. This paper presents a methodology using LightGBM classifier with hyper parameter tuned using Grasshopper Optimization Algorithm using NetFlow based UNSW-NB15 dataset. The proposed methodology aims to give a contribution to several works done using NetFlow based data, achieving ∼97% of the detection rate.