Unsupervised Learning of Debiased Representations with Pseudo-Attributes

TL;DR

This paper introduces an unsupervised method to learn debiased representations by identifying pseudo-attributes through clustering, improving fairness and robustness without requiring explicit bias annotations.

Contribution

It proposes a novel unsupervised debiasing technique using clustering and cluster-wise reweighting, eliminating the need for human bias annotations.

Findings

Outperforms existing methods on multiple benchmarks.

Achieves competitive accuracy with supervised approaches.

Effectively mitigates dataset bias and improves worst-case generalization.

Abstract

Dataset bias is a critical challenge in machine learning since it often leads to a negative impact on a model due to the unintended decision rules captured by spurious correlations. Although existing works often handle this issue based on human supervision, the availability of the proper annotations is impractical and even unrealistic. To better tackle the limitation, we propose a simple but effective unsupervised debiasing technique. Specifically, we first identify pseudo-attributes based on the results from clustering performed in the feature embedding space even without an explicit bias attribute supervision. Then, we employ a novel cluster-wise reweighting scheme to learn debiased representation; the proposed method prevents minority groups from being discounted for minimizing the overall loss, which is desirable for worst-case generalization. The extensive experiments demonstrate the outstanding performance of our approach on multiple standard benchmarks, even achieving the competitive accuracy to the supervised counterpart.