PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

A mixture model-based on-line CEM algorithm
Allou Same, Gérard Govaert and Christophe Ambroise
In: Advances in Intelligent Data Analysis, 6th International Symposium on Data Analysis, IDA 2005, 8-10 Oct 2005, Madrid, Spain.

Abstract

he classification EM algorithm (CEM), applied using mixture models is a very useful clustering algorithm which generalize the well known k-means algorithm. This algorithm, in a practical point of view, is faster that EM algorithm and converges in a few iterations. Many actual applications require massive data sets to be classified in a real time. In that context, CEM algorithm has been used in our application dealing with a real-time flaw diagnosis for pressurized containers using acoustic emissions. Particularly, CEM algorithm has been used for the clustering of acoustic emissions whose size increase in the time. However, when the number of acoustic emissions is more than 10000, CEM algorithm has not been able to react in real time. In this work, we aim to develop an on-line mixture model based clustering algorithm.

EPrint Type:Conference or Workshop Item (Talk)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Computational, Information-Theoretic Learning with Statistics
ID Code:1954
Deposited By:Gérard Govaert
Deposited On:30 December 2005