Semi-Supervised Learning of Class Balance under Class-Prior Change by Distribution Matching

TL;DR

This paper introduces a method to estimate class ratios in test datasets with unknown class balance by matching input data distributions, addressing bias issues in real-world classification tasks.

Contribution

It proposes a novel distribution matching approach for class ratio estimation under class-prior change without requiring labeled test data.

Findings

Effective class ratio estimation demonstrated in experiments

Reduces bias caused by class imbalance in test data

Applicable to real-world classification scenarios

Abstract

In real-world classification problems, the class balance in the training dataset does not necessarily reflect that of the test dataset, which can cause significant estimation bias. If the class ratio of the test dataset is known, instance re-weighting or resampling allows systematical bias correction. However, learning the class ratio of the test dataset is challenging when no labeled data is available from the test domain. In this paper, we propose to estimate the class ratio in the test dataset by matching probability distributions of training and test input data. We demonstrate the utility of the proposed approach through experiments.