Friday 22 February 2008

Multi-Agent Reinforcement Learning for Intrusion Detection: A case study and evaluation

I gave this seminar on Feb 22nd.
Artificial Intelligence Group. Computer Sciences, University of York
Abstract:

In this seminar I will present an architecture of distributed sensor and decision agents that learn how to identify normal and abnormal states of the network using Reinforcement Learning (RL). Sensor agents extract network state information using tile-coding as a function approximation technique and send communication signals in the form of actions to decision agents. These in turn generate actions in the form of alarms to the network operator. By means of an on-line process, sensor and decision agents learn the semantics of the communication actions without any previous knowledge. In this presentation I will describe the learning process, the operation of the agent architecture and the evaluation results of our research work.

The presentation is here:




And a video of a Denial of Service Attack. Disclaimer: It may be disturbing for certain audience (it contains cheesy music from ABBA)