Nearest Neighbor-based Importance Weighting

TL;DR

This paper introduces a simple, effective importance weighting method based on nearest neighbor classification, which addresses covariate shift problems and performs well across various tasks.

Contribution

It presents a novel importance weighting approach using nearest neighbor classification, offering a straightforward and effective alternative to existing methods.

Findings

Demonstrates effectiveness of NNeW in classification tasks

Performs well as a baseline method for importance weighting

Shows competitive results in covariate shift scenarios

Abstract

Importance weighting is widely applicable in machine learning in general and in techniques dealing with data covariate shift problems in particular. A novel, direct approach to determine such importance weighting is presented. It relies on a nearest neighbor classification scheme and is relatively straightforward to implement. Comparative experiments on various classification tasks demonstrate the effectiveness of our so-called nearest neighbor weighting (NNeW) scheme. Considering its performance, our procedure can act as a simple and effective baseline method for importance weighting.