Abstract: As of late, with wide utilization of PC frameworks, web, and fast development of PC systems, the issue of interruption discovery in system security has turned into a critical issue of concern. In such manner, different interruption recognition frameworks have been produced for utilizing abuse recognition and inconsistency discovery strategies. These frameworks attempt to make strides identification rates of variety in assault sorts and diminish the false positive rate. In this paper, another interruption discovery strategy has been presented utilizing Min Max K-means clustering algorithm, which defeats the lack of affectability to starting focuses in K-means algorithm, and expands the nature of clustering. The investigates the NSL-KDD information set demonstrate that the proposed strategy is more effective than that in view of K-means clustering algorithm. Additionally, the strategy has higher discovery rate and lower false positive recognition rate.

Keywords: K-means algorithm, NSL-KDD, clustering algorithm.