Fair Classification under Covariate Shift and Missing Protected Attribute -- an Investigation using Related Features

TL;DR

This paper explores fair classification in scenarios with covariate shift and missing protected attributes, proposing a method that combines importance-weighting and related features to improve fairness and accuracy.

Contribution

It introduces a novel approach that integrates importance-weights and related features to address fairness issues under covariate shift and missing protected attributes.

Findings

Effectively handles covariate shift with importance-weights

Improves fairness in classification with missing protected attributes

Demonstrates the approach's effectiveness on relevant datasets

Abstract

This study investigated the problem of fair classification under Covariate Shift and missing protected attribute using a simple approach based on the use of importance-weights to handle covariate-shift and, Related Features arXiv:2104.14537 to handle missing protected attribute.