Radius kNN Classifier Using Aggregation of Fuzzy Equivalences

Abstract

The paper presents a modified classification method based on the k-nearest neighbor algorithm. In the modified kNN algorithm some aggregations of fuzzy equivalences are used instead of metrics and the selection of the nearest neighbors is limited by their closeness from a tested object. This procedure is intended to improve suitability of the kNN algorithm, when a significant part of the closest neighbors is not close enough to the tested object. Additionally, some theoretical results concerning fuzzy equivalences and their aggregations are included in the paper.