Intrusion detection with comparative analysis of supervised learning techniques and fisher score feature selection algorithm

Abstract

Rapid development of technologies not only makes life easier, but also reveals a lot of security problems. Developing and changing of attack types affect many people, organizations, companies etc. Therefore, intrusion detection systems have been developed to avoid financial and emotional loses. In this paper, we used CICIDS2017 dataset which consist of benign and the most cutting-edge common attacks. Best features are selected by using Fisher Score algorithm. Real world data extracted from the dataset are classified as DDoS or benign with using Support Vector Machine (SVM), K Nearest Neighbour (KNN) and Decision Tree (DT) algorithms. As a result of the study, 0,9997%, 0,5776%, 0,99% success rates were achieved respectively. © Springer Nature Switzerland AG 2018.