2026-06-24T11:47:58Z https://meral.edu.mm/oai

oai:meral.edu.mm:recid/3616 2021-12-13T00:34:37Z 1582963302567:1597824273898 user-ucsy

Effective Anomaly Detection Using Hidden-Semi Markov Model Wutyi, Khaing Shwe Thwin, Mie Mie Su Anomaly detection studies the normal behaviorof the monitored system and then looks out for anydifference in it to detect anomalies or attacks. It is ableto detect new attacks as any attack is assumed to bedifferent from normal activity. It sometimes sets falsealarms because it erroneously classifies the normaluser behaviors as attacks.Different techniques havebeen used for anomaly detector generation.In thispaper, we would like to propose Hidden-Semi MarkovModel (HSMM) as it is introduced in intrusiondetection for several years. Based on this HSMM, analgorithm of anomaly detection is presented in thispaper, which computes the distance between theprocesses monitored by intrusion detection system andthe perfect normal processes. In this algorithm, we usethe average information entropy (AIE) of fixed-lengthobserved sequence as the anomaly detection metricbased on maximum entropy principle (MEP). Toimprove accuracy, the segmental K-means algorithm isapplied as training algorithm for the HSMM. Bycomparing the accurate rate with the experimentalresults of previous research, it shows that our methodcan perform a more accurate detection. 2015-02-05 http://hdl.handle.net/20.500.12678/0000003616 https://meral.edu.mm/records/3616