Bayesian Model Selection Methods for Mutual and Symmetric $k$-Nearest Neighbor Classification
Abstract
The -nearest neighbor classification method (-NNC) is one of the simplest nonparametric classification methods. The mutual -NN classification method (MNNC) is a variant of -NNC based on mutual neighborship. We propose another variant of -NNC, the symmetric -NN classification method (SNNC) based on both mutual neighborship and one-sided neighborship. The performance of MNNC and SNNC depends on the parameter as the one of -NNC does. We propose the ways how MNN and SNN classification can be performed based on Bayesian mutual and symmetric -NN regression methods with the selection schemes for the parameter . Bayesian mutual and symmetric -NN regression methods are based on Gaussian process models, and it turns out that they can do MNN and SNN classification with new encodings of target values (class labels). The simulation results show that the proposed methods are better than or comparable to -NNC, MNNC and SNNC with the parameter 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