Data Debiasing with Datamodels (D3M): Improving Subgroup Robustness via Data Selection

TL;DR

This paper introduces D3M, a data debiasing method that selectively removes training examples causing subgroup failures, improving model robustness without needing group labels or extensive tuning.

Contribution

D3M is a novel data selection technique that isolates and removes problematic examples to enhance subgroup robustness without requiring group annotations.

Findings

Effective in improving subgroup performance

Removes minimal data points for debiasing

Does not need group labels or hyperparameter tuning

Abstract

Machine learning models can fail on subgroups that are underrepresented during training. While techniques such as dataset balancing can improve performance on underperforming groups, they require access to training group annotations and can end up removing large portions of the dataset. In this paper, we introduce Data Debiasing with Datamodels (D3M), a debiasing approach which isolates and removes specific training examples that drive the model's failures on minority groups. Our approach enables us to efficiently train debiased classifiers while removing only a small number of examples, and does not require training group annotations or additional hyperparameter tuning.