Fine-Grained Class-Conditional Distribution Balancing for Debiased Learning

TL;DR

This paper introduces FG-CCDB, a fine-grained distribution matching method for debiased learning, leveraging a multi-stage re-training strategy to better address spurious correlations without bias annotations.

Contribution

It proposes FG-CCDB, a novel fine-grained distribution matching technique that improves debiased learning by using confusion-cell-wise reweighting and a multi-stage re-training strategy.

Findings

FG-CCDB outperforms bias-supervised methods in multi-class and shortcut scenarios.

MST effectively approximates ground-truth bias annotations.

FG-CCDB achieves comparable results to bias-supervised approaches in binary classification.

Abstract

Achieving group-robust generalization in the presence of spurious correlations remains a significant challenge, particularly when bias annotations are unavailable. Recent studies on Class-Conditional Distribution Balancing (CCDB) reveal that spurious correlations often stem from mismatches between the class-conditional and marginal distributions of bias attributes. They achieve promising results by addressing this issue through simple distribution matching in a bias-agnostic manner. However, CCDB approximates each distribution using a single Gaussian, which is overly simplistic and rarely holds in real-world applications. To address this limitation, we propose a novel Multi-stage data-Selective reTraining strategy (MST), which describes each distribution in greater detail using the hard confusion matrix. Building on these finer descriptions, we propose a fine-grained variant of CCDB, termed FG-CCDB, which enhances distribution matching through more precise confusion-cell-wise reweighting. FG-CCDB learns sample weights from a global perspective, effectively mitigating spurious correlations without incurring substantial storage or computational overhead. Extensive experiments demonstrate that MST serves as a reliable proxy for ground-truth bias annotations and can be seamlessly integrated with bias-supervised methods. Moreover, when combined with FG-CCDB, our method performs on par with bias-supervised approaches on binary classification tasks and significantly outperforms them in highly biased multi-class and multi-shortcut scenarios.