Abstract
In real-time, Network anomaly-based intrusion detection systems countenance the defiance of noticing the novel anomalies. In this paper we present network anomaly detection system with adaptive outlier detection approach which based on frequent patterns technique. The significant advantage of the proposed approach lay in three aspects: effective and straightforward to process online traffic data; adaption to the changes in the traffic streams; the ability to detect the anomalous once it occurs. The experiments indicate a good detection for the new anomalous behavior and the performance of our proposed approach is approximately near to the static approach.