An Ensemble Machine Learning Botnet Detection Framework Based on Noise Filtering

Abstract

During the past decade, one of the most serious cyber threats has been the growth of botnet. Since botnet attacks combine the characteristics of many malicious attacks, they have complex attack behaviors and communication patterns. In order to improve the detection rate, many researchers use machine learning techniques. In this paper, we proposed an ensemble classification framework based on noise filtering to improve detection performance. The experimental results show that the proposed framework improves the detection rate and reduces the false alarm rate. We also compare the proposed classification model with other ensemble classification models. The experimental results also show that the classification model has the highest accuracy and lower false alarm rate.