The objective of the Phase I effort is to produce a "proof of concept". The goal is to demonstrate that automatically learned rules generated by Meta-Learning Agent Technology can use data mining techniques to generate efficient N-Code for real-time use with the Network Flight Recorder Intrusion Detection System. Columbia University's JAM Project software provides the models. This architecture will enable rapid development and deployment of learned, network based intrusion detection models. The goal would be to demonstrate the results of the research in DARPA's YR2000 real-time intrusion detection evaluation. This effort advances the security research in anomaly-based intrusion detection.
Keywords: Network Intrusion Detection; Host Based Ids; Meta-Learning
