%0 Journal Article
%T An Anomaly Detection Algorithm Based on Hidden Pattern<br>基于隐含模式的异常检测算法
%A Xiang Kui
%A Jiang Jing-ping
%A <br>向馗
%A 蒋静坪
%J 电子与信息学报
%D 2007
%I 
%X It is a difficult problem how to detect such accident of a system. This paper presents a new algorithm, an anomaly detection algorithm based on hidden pattern. Epsilon machine, a new computational mechanics, can discover hidden pattern from the response time series. Causal State Splitting Reconstruction (CSSR), one algorithm of epsilon machine, can infer a set of causal states, which has an analogy to hidden Markov chain. Based on this set, an anomaly measure can be defined, which is the distance of two characteristic vectors. Computing all parts of the time series, an anomaly evolution curve can be got. In simulation analysis of Duffing equation, step changes appear in the anomaly curve, before Duffing oscillator begin to bifurcate. The algorithm proves to be effective in anomaly detection and warning.
%K Anomaly detection
%K Hidden pattern
%K Epsilon machine
%K Time series<br>异常检测
%K 隐含模式
%K ε机
%K 时间序列
%U http://www.alljournals.cn/get_abstract_url.aspx?pcid=5B3AB970F71A803DEACDC0559115BFCF0A068CD97DD29835&cid=1319827C0C74AAE8D654BEA21B7F54D3&jid=EFC0377B03BD8D0EF4BBB548AC5F739A&aid=B36CFA6E52098E16&yid=A732AF04DDA03BB3&vid=771469D9D58C34FF&iid=B31275AF3241DB2D&sid=BFA4330C9764AE1A&eid=CBBCA10F35A35A94&journal_id=1009-5896&journal_name=电子与信息学报&referenced_num=0&reference_num=8