Une méthode possibiliste de discrimination adaptée aux classes de forme complexe

Mathématiques et sciences humaines. Mathematics and social sciences

Billaudel, Patrice¹  Villermain Lecolier, Gérard¹  Devillez, Arnaud¹ 

关键词: classification;    fuzzy logic;    possibility theory;    multicriteria decision;   

DOI  :  10.4000/msh.2800

学科分类：数学（综合）

来源: College de France * Ecole des Hautes Etudes en Sciences Sociales (E H E S S)

【 摘 要 】

Our team works on the classification of data coming from industrial and medical sectors, in order to develop decision making and diagnosis systems. In this paper we propose to modify the fuzzy method of pattern matching, in order to classify data including classes of complex shape. We describe the basic method before showing its limits when classes are not convex. Then, we propose to improve the method by introducing a multiprototype approach. We present an industrial example, which consists in sorting automatically plastic bottles in order to recycle them. Finally, we compare the results obtained by this method with those given by the fuzzy k-nearest neighbours method, using three types of data : plastic bottles, iris and waveform data.

【 授权许可】

Unknown