Enhancing Intrinsic Features for Debiasing via Investigating Class-Discerning Common Attributes in Bias-Contrastive Pair

TL;DR

This paper introduces a novel debiasing method for image classification that explicitly guides models to focus on intrinsic features by investigating common attributes in bias-contrastive pairs, leading to improved performance.

Contribution

It proposes a new approach that identifies and enhances intrinsic features through class-discerning common attributes without relying on bias labels, improving debiasing effectiveness.

Findings

Achieves state-of-the-art results on synthetic and real-world datasets.

Effectively guides models to focus on intrinsic features.

Improves robustness against dataset bias.

Abstract

In the image classification task, deep neural networks frequently rely on bias attributes that are spuriously correlated with a target class in the presence of dataset bias, resulting in degraded performance when applied to data without bias attributes. The task of debiasing aims to compel classifiers to learn intrinsic attributes that inherently define a target class rather than focusing on bias attributes. While recent approaches mainly focus on emphasizing the learning of data samples without bias attributes (i.e., bias-conflicting samples) compared to samples with bias attributes (i.e., bias-aligned samples), they fall short of directly guiding models where to focus for learning intrinsic features. To address this limitation, this paper proposes a method that provides the model with explicit spatial guidance that indicates the region of intrinsic features. We first identify the intrinsic features by investigating the class-discerning common features between a bias-aligned (BA) sample and a bias-conflicting (BC) sample (i.e., bias-contrastive pair). Next, we enhance the intrinsic features in the BA sample that are relatively under-exploited for prediction compared to the BC sample. To construct the bias-contrastive pair without using bias information, we introduce a bias-negative score that distinguishes BC samples from BA samples employing a biased model. The experiments demonstrate that our method achieves state-of-the-art performance on synthetic and real-world datasets with various levels of bias severity.