BPkNN: K-Nearest Neighbor Classifier with Pairwise Distance Metrics and Belief Function Theory

Lianmeng Jiao, Xiaojiao Geng, Quan Pan

Research output: Contribution to journalArticlepeer-review

26 Scopus citations

Abstract

The k -nearest neighbor (kNN) rule is one of the most popular classification algorithms in pattern recognition field because it is very simple to understand but works quite well in practice. However, the performance of the k NN rule depends critically on its being given a good distance metric over the input space, especially in small data set situations. In this paper, a new k NN-based classifier, called BP k NN, is developed based on pairwise distance metrics and belief function theory. The idea of the proposal is that instead of learning a global distance metric, we first decompose it into learning a group of pairwise distance metrics. Then, based on each learned pairwise distance metric, a pairwise k NN ( \text{P}k NN) sub-classifier can be adaptively designed to separate two classes. Finally, a polychotomous classification problem is solved by combining the outputs of these \text{P}k NN sub-classifiers in belief function framework. The BP k NN classifier improves the classification performance thanks to the new distance metrics which provide more flexibility to design the feature weights and the belief function-based combination method which can better address the uncertainty involved in the outputs of the sub-classifiers. Experimental results based on synthetic and real data sets show that the proposed BP k NN can achieve better classification accuracy in comparison with some state-of-the-art methods.

Original languageEnglish
Article number6287639
Pages (from-to)48935-48947
Number of pages13
JournalIEEE Access
Volume7
DOIs
StatePublished - 2019

Keywords

  • Belief function theory
  • K-nearest-neighbor classifier
  • Pairwise distance metric
  • Pattern classification

Fingerprint

Dive into the research topics of 'BPkNN: K-Nearest Neighbor Classifier with Pairwise Distance Metrics and Belief Function Theory'. Together they form a unique fingerprint.

Cite this