Performance Enhancement of Intrusion Detection System Using Neural Network Technique

Security issues, such as network intrusion and virus infection, are becoming more and more serious with the growth of computer and network applications. Intrusion is a set of actions which attempt to compromise the confidentiality, integrity or availability of a resources Intrusion detection systems are used to monitor computer system for sign of security violations. On detection of miscellaneous intrusion from the World Wide Web, we need effective intrusion detection system. In Practice, IDSs have been observed to trigger thousands of alerts per day, most of which are mistakenly triggered by begin events such as false positive. To address the problem of false positives clustering approach is used to groups similar type of attacks to enhance the performance of IDS. Here in this paper we present and implemented two clustering algorithm K-means centroid based data mining algorithm and Neural gas competive hebbian learning approach. These two algorithms are applied on intrusion detection dataset i.e KDDCUP99 dataset. The obtained results of Neural gas competive hebbian learning approach performs better in terms of Sum of square error (SSE).