Modality-Agnostic Debiasing for Single Domain Generalization

TL;DR

This paper introduces a versatile, modality-agnostic debiasing framework called MAD that enhances single domain generalization across various data modalities by disentangling domain-specific and domain-general features.

Contribution

MAD is a novel, pluggable two-branch classifier framework that improves single-DG performance across multiple modalities, including text, images, and 3D point clouds.

Findings

MAD improves accuracy on 3D point clouds by 2.82%.

MAD increases mIOU on semantic segmentation by 1.5%.

MAD outperforms existing methods in diverse modality scenarios.

Abstract

Deep neural networks (DNNs) usually fail to generalize well to outside of distribution (OOD) data, especially in the extreme case of single domain generalization (single-DG) that transfers DNNs from single domain to multiple unseen domains. Existing single-DG techniques commonly devise various data-augmentation algorithms, and remould the multi-source domain generalization methodology to learn domain-generalized (semantic) features. Nevertheless, these methods are typically modality-specific, thereby being only applicable to one single modality (e.g., image). In contrast, we target a versatile Modality-Agnostic Debiasing (MAD) framework for single-DG, that enables generalization for different modalities. Technically, MAD introduces a novel two-branch classifier: a biased-branch encourages the classifier to identify the domain-specific (superficial) features, and a general-branch captures domain-generalized features based on the knowledge from biased-branch. Our MAD is appealing in view that it is pluggable to most single-DG models. We validate the superiority of our MAD in a variety of single-DG scenarios with different modalities, including recognition on 1D texts, 2D images, 3D point clouds, and semantic segmentation on 2D images. More remarkably, for recognition on 3D point clouds and semantic segmentation on 2D images, MAD improves DSU by 2.82\% and 1.5\% in accuracy and mIOU.