A Density-based Undersampling Approach to Intrusion Detection
Mahsa Mir Mohseni, Jafar Tanha · 2021
In this paper, we consider the intrusion detection task using machine learning approach. An intrusion detection system mainly detects and reports suspicious activities even after passing through the firewall or other security equipment. The main issue in this task is that the data is originally imbalanced and the traditional learning algorithms may not discover proper patterns from the minority class. We propose a cluster-based undersampling approach using a density-based clustering approach to learn from the CICIDS2017 dataset. We further propose a novel measurement to sample a set of representative data points from the majority class. Our experimental results indicate that our proposed algorithm performs better than the state-of-the-art methods