Event and anomaly detection using Tucker3 decomposition

Workshop on Ubiquitous Data Mining 2012

计算机科学

Others  :  http://ceur-ws.org/Vol-960/paper2.pdf
PID  :  28234

学科分类：计算机科学（综合）

来源: CEUR

【 摘 要 】

1 Failure detection in telecommunication networks is a vital task. So far, several supervised and unsupervised solutionshave been provided for discovering failures in such networks.Among them unsupervised approaches has attracted more attentionsince no label data is required [1]. Often, network devices are notable to provide information about the type of failure. In such cases,unsupervised setting is more appropriate for diagnosis. Amongunsupervised approaches, Principal Component Analysis (PCA)has been widely used for anomaly detection literature and can beapplied to matrix data (e.g. Users-Features). However, one of theimportant properties of network data is their temporal sequentialnature. So considering the interaction of dimensions over a thirddimension, such as time, may provide us better insights into thenature of network failures. In this paper we demonstrate the powerof three-way analysis to detect events and anomalies in time evolving network data.