Events — Colloquia & Seminars
Statistical Models for Anomaly Detection
Speaker: Alvaro Cardenas, University of California, Berkeley
Date: Wednesday, April 22, 2009
Talk: 11:00 AM, 366 WVH
Abstract
In this talk I will present new statistical models for different problems in computer security. The first half of the talk will focus on network anomaly detection. I will show how statistical methods and game theory can help practitioners understand and design better anomaly detection schemes in the face of intelligent and adaptive attackers. I will also describe the Bayesian Receiver Operating Characteristic (B-ROC) curves, and I will explain how they can be used as a tool to interpret the classification accuracy of intrusion detection systems.
In the second half of the talk I will present some of my ongoing work in two different fields: the cybercrime ecosystem and the security of cyber-physical systems. In particular, I will show how to use statistical techniques to extract information and model parts of the computer underground market. I will also show how a mathematical model of the physical world can be used to detect computer attacks against cyber-physical systems, and to design attack-resilient control algorithms.
Brief Biography
Alvaro Cardenas is a postdoctoral scholar at the University of California, Berkeley. He received his MS and Ph.D. degrees from the University of Maryland, College Park, and his B.S. from the Universidad de los Andes, Bogota (Colombia). His research interests include intrusion detection systems, the security of control systems, cybercrime, and the applications of probability, game theory, optimization and machine learning to these fields. He received a two-year graduate school fellowship from the University of Maryland and a two-year distinguished research assistantship from the Institute of Systems Research.