Unsupervised Learning for Intrusion Detection Using L2E

• Activity Number: 402
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Umashanger Thayasivam*+
• Companies: Rowan University
• Keywords: Intrusion detection ; Unsupervised ; L2E ; GMM ; Supervised ; Classification

Abstract

Application and development of specialized machine learning techniques is gaining increasing attention in the intrusion detection community. A variety of learning techniques proposed for different intrusion detection problems can be roughly classified into two broad categories: supervised (classification) and unsupervised (anomaly detection and clustering). Intrusion detection is conducted by adopting Gaussian mixture models (GMM), an unsupervised learning technique. In this contribution we develop an experimental framework for GMM parameter estimation using a novel approach with minimizing integrated square distance (L2E). We then investigate both kind of learning(supervised & Unsupervised) techniques including our L2E, with respect to their detection accuracy and ability to detect unknown attacks.