Multiview Adaptive K-Nearest Neighbor Classification

Abstract

The k-nearest neighbor (KNN) classifies unlabeled samples according to the parameter <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$k$</tex-math></inline-formula> , which is a user-defined constant and usually depends on prior knowledge. The selection of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$k$</tex-math></inline-formula> is crucial, as the size of the sample neighborhood affects the classification accuracy. To tackle this issue, we introduce the adaptive KNN (AKNN), which constructs a decision tree to assign different numbers of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$k$</tex-math></inline-formula> -values to different samples. In AKNN, we use the sample label information to calculate the weight between samples. Furthermore, to extend AKNN to a multiview scenario, we propose a method namely multiview adaptive KNN (MVAKNN), which integrates information from every single view by using the Dempster–Shafer theory. We conduct experiments on three benchmark multiview image datasets and the results show that MVAKNN exhibits desirable classification accuracy, outperforming some single-view and multiview methods. Experiments with Gaussian noises show the robustness of the proposed method.