Designing a technique for detecting intrusion based on modified Adaptive Resonance Theory Network

Detecting intrusion is an important activity in the solution finding algorithms. Artificial Neural Networks can be employed for the detection of intrusion. This paper discusses a technique for detecting intrusion by modified ART network (Adaptive Resonance Theory Network). The various states associated with an event are processed to frame a basic model for detection. The probabilities associated with these states are utilized to define a mathematical function. This function determines or evaluates the performance of the system. The basic ART network is modified using the decisions based on derivative this function. The design of modified ART networks is described. Thus this paper shows the effectiveness of modified ART network in detecting intrusion.