Teaching Computers to Catch Hackers

No ratings

Presented at BSidesLondon 2010 by

Two exciting fields in computer science are those of machine learning and computer security. The speaker finds both of these rather interesting, and would like to convince you of the same. This talk will look at clever techniques to teach computers to detect malicious activity on computing infrastructure i.e. intrusion detection systems (IDS). The speaker has conducted research on applying machine learning and artificial intelligence techniques to detecting malicious TCP traffic. He will explain where he got the 20GB of traffic to use, and how he used freely available tools (and a few Perl scripts) to teach a computer to detect a high percentage of hacker-like activity. He will also explain the shortcomings of his research, and throw out possible further work to the audience to get involved in. The research was part of his MSc in computer science from the University of Oxford.