English

Bayesian Model Selection Methods for Mutual and Symmetric $k$-Nearest Neighbor Classification

Machine Learning 2016-08-16 v1 Machine Learning

Abstract

The kk-nearest neighbor classification method (kk-NNC) is one of the simplest nonparametric classification methods. The mutual kk-NN classification method (MkkNNC) is a variant of kk-NNC based on mutual neighborship. We propose another variant of kk-NNC, the symmetric kk-NN classification method (SkkNNC) based on both mutual neighborship and one-sided neighborship. The performance of MkkNNC and SkkNNC depends on the parameter kk as the one of kk-NNC does. We propose the ways how MkkNN and SkkNN classification can be performed based on Bayesian mutual and symmetric kk-NN regression methods with the selection schemes for the parameter kk. Bayesian mutual and symmetric kk-NN regression methods are based on Gaussian process models, and it turns out that they can do MkkNN and SkkNN classification with new encodings of target values (class labels). The simulation results show that the proposed methods are better than or comparable to kk-NNC, MkkNNC and SkkNNC with the parameter kk selected by the leave-one-out cross validation method not only for an artificial data set but also for real world data sets.

Cite

@article{arxiv.1608.04063,
  title  = {Bayesian Model Selection Methods for Mutual and Symmetric $k$-Nearest Neighbor Classification},
  author = {Hyun-Chul Kim},
  journal= {arXiv preprint arXiv:1608.04063},
  year   = {2016}
}

Comments

8 pages

R2 v1 2026-06-22T15:19:19.033Z