Learning to Generate Novel Domains for Domain Generalization

TL;DR

This paper introduces L2A-OT, a novel data augmentation method using optimal transport to synthesize diverse pseudo-novel domains, significantly improving model generalization in domain generalization tasks.

Contribution

The paper proposes a new data augmentation approach for domain generalization that synthesizes diverse domains via optimal transport, enhancing generalization performance.

Findings

L2A-OT outperforms state-of-the-art DG methods on four benchmarks.

Synthesizing pseudo-novel domains increases training data diversity.

Optimal transport effectively models divergence between source and synthetic domains.

Abstract

This paper focuses on domain generalization (DG), the task of learning from multiple source domains a model that generalizes well to unseen domains. A main challenge for DG is that the available source domains often exhibit limited diversity, hampering the model's ability to learn to generalize. We therefore employ a data generator to synthesize data from pseudo-novel domains to augment the source domains. This explicitly increases the diversity of available training domains and leads to a more generalizable model. To train the generator, we model the distribution divergence between source and synthesized pseudo-novel domains using optimal transport, and maximize the divergence. To ensure that semantics are preserved in the synthesized data, we further impose cycle-consistency and classification losses on the generator. Our method, L2A-OT (Learning to Augment by Optimal Transport) outperforms current state-of-the-art DG methods on four benchmark datasets.