Hyper Parameter Optimized NIDS via Machine Learning in IoT Ecosystem

Today’s dynamic and hyperactive world is confined to a ball size connectivity only via network and internet. People knowing or unknowingly risk their privacy at a huge rate in the network. Undoubtedly, this is the demand of the time to be digitized and technical. As the population increases, the threat to security increases as well. To detect cyber-attacks and malicious threats to the network, network intrusion detection system is a very prominent concept to deal with them. There is no standard protocol to build up NIDS from scratch as it varies accordingly. Machine learning based NIDS paved a way towards autonomous working of this system keeping in mind that the systems can’t be compared, not even on basic entities or architecture of the system. The paper introduces the recent work and circumstances of the NIDS though the authors didn’t consider the working technique of all the works while implementing. Secondly, a proposed solution for the detection and its methodology has been described in detail while testing it on the UGR ‘16 dataset for further evaluation and benchmarking the durability and efficiency of the model.