Unsupervised Clustering in an Orthogonal System

IPSJ

Search this article

Description

It is known that the Karhunen-Loeve orthogonal system provides a convenient tool to unsupervised clustering problem.Though some results have been obtained along this line the method is not yet fully applied to the practical pattern clustering problem.The method of unsupervised clustering problem is first explained from the K-L system point of view.Secondly an algorithm is proposed for the unsupervised clustering problem.Finally the algorithm is applied to reformed patterns which are obtained from practical patterns by the method of reformation proposed by the author.Computersimulated experiments are carried into effect in order to test the usefullness of the present method.

It is known that the Karhunen-Loeve orthogonal system provides a convenient tool to unsupervised clustering problem.Though some results have been obtained along this line,the method is not yet fully applied to the practical pattern clustering problem.The method of unsupervised clustering problem is first explained from the K-L system point of view.Secondly,an algorithm is proposed for the unsupervised clustering problem.Finally,the algorithm is applied to reformed patterns,which are obtained from practical patterns by the method of reformation proposed by the author.Computersimulated experiments are carried into effect in order to test the usefullness of the present method.

Journal

Details 詳細情報について

Report a problem

Back to top