ProxiMix: Enhancing Fairness with Proximity Samples in Subgroups

TL;DR

ProxiMix is a novel data augmentation method that improves fairness in machine learning by maintaining proximity relationships during label generation, effectively reducing biases across multiple datasets and models.

Contribution

It introduces ProxiMix, a new pre-processing strategy combining mixup and proximity-aware label generation to enhance fairness in ML models.

Findings

ProxiMix improves fairness in predictions.

ProxiMix enhances fairness of recourse.

Effective across various datasets and models.

Abstract

Many bias mitigation methods have been developed for addressing fairness issues in machine learning. We found that using linear mixup alone, a data augmentation technique, for bias mitigation, can still retain biases present in dataset labels. Research presented in this paper aims to address this issue by proposing a novel pre-processing strategy in which both an existing mixup method and our new bias mitigation algorithm can be utilized to improve the generation of labels of augmented samples, which are proximity aware. Specifically, we proposed ProxiMix which keeps both pairwise and proximity relationships for fairer data augmentation. We conducted thorough experiments with three datasets, three ML models, and different hyperparameters settings. Our experimental results showed the effectiveness of ProxiMix from both fairness of predictions and fairness of recourse perspectives.