Outlier-Aware Training for Improving Group Accuracy Disparities

TL;DR

This paper introduces an outlier detection method to improve group fairness in training by removing unlearnable examples, leading to better accuracy and error detection compared to existing reweighting techniques.

Contribution

It proposes a novel outlier detection approach to enhance group fairness training, addressing issues with unlearnable examples in reweighting methods.

Findings

Achieves competitive or superior accuracy compared to JTT.

Effectively detects and removes annotation errors.

Improves model fairness by mitigating unlearnable examples.

Abstract

Methods addressing spurious correlations such as Just Train Twice (JTT, arXiv:2107.09044v2) involve reweighting a subset of the training set to maximize the worst-group accuracy. However, the reweighted set of examples may potentially contain unlearnable examples that hamper the model's learning. We propose mitigating this by detecting outliers to the training set and removing them before reweighting. Our experiments show that our method achieves competitive or better accuracy compared with JTT and can detect and remove annotation errors in the subset being reweighted in JTT.