An induction learning approach for building intrusion detection models using genetic algorithms

Building and updating an effective intrusion detection system is complex engineering knowledge. A method of learning the intrusion detection rules based on Genetic Algorithms is presented in order to realize the automation of the detection models. The same attributes of an intrusion can be found through the heuristic search in the network data space. In our experiments the characters of an attack, such as smurf, are summarized inductively through the datasets of the 1998 DARPA Intrusion Detection Evaluation Program. The effectiveness and robustness of the approach are proved.