Semi-supervised Botnet Detection Using Ant Colony Clustering
Khalid Huseynov, Kwangjo Kim, Paul D. Yoo · Scandinavian Conference on Information Systems · 2014
Recently, botnets have become one of the fast growing and changing vectors of malicious underground economy. They pose serious threats on the cyber-security of citizens, enterprises, and governments. Many recent countermeasures utilize machine-learning techniques due to its adaptability and model-free properties. In this research, we propose a bio-inspired computing technique called ant colony clustering for the accurate, scalable detection of botnet attacks. The proposed method is able to detect the botnet hosts rapidly and accurately while not depending on its traffic payload. Furthermore, it utilizes only a small sample of labeled data in the form of semi-supervised learning.