arXiv.itsnick.co

an independent field guide to 23 million papers from arXiv & OpenAlex — not affiliated with arXiv.org

Social Sciences · Urban Studies

Generalizing the Evidence-Theoretic k-NN rule to Fuzzy Pattern Recognition

L. M. Zouhal, T. Denœux

See on the map

Abstract

A pattern classifier using neighborhood information in the framework of Dempster-Shafer theory was recently proposed. This method consists in considering each neighbor of a pattern x to be classified as an item of evidence supporting certain hypotheses concerning the class membership of x. This information is represented by a belief structure defined as a function of the distance between the two vectors. The evidence of the k nearest neighbors is then pooled by means of Dempster's rule of combination. In this paper, this approach is extended to handle the more general situation in which training patterns are assumed to have some degree of membership to each class. Two methods are described. The first one exploits the relationship between fuzzy sets, possibility distributions and consonant belief structures. The second approach is based on a generalization of the Dempster-Shafer theory to fuzzy sets. These methods are compared experimentally with the voting, evidence-theoretic and fuzzy...

More like this