Learning to Generalize: Meta-Learning for Domain Generalization

TL;DR

This paper introduces a meta-learning approach for domain generalization that trains models to perform well on unseen domains by simulating domain shifts during training, achieving state-of-the-art results in image classification and reinforcement learning.

Contribution

It proposes a model-agnostic meta-learning method that synthesizes virtual testing domains within each mini-batch to improve generalization to new domains.

Findings

Achieved state-of-the-art results on cross-domain image classification.

Demonstrated effectiveness on reinforcement learning tasks.

Validated the approach's potential for various domain shift scenarios.

Abstract

Domain shift refers to the well known problem that a model trained in one source domain performs poorly when applied to a target domain with different statistics. {Domain Generalization} (DG) techniques attempt to alleviate this issue by producing models which by design generalize well to novel testing domains. We propose a novel {meta-learning} method for domain generalization. Rather than designing a specific model that is robust to domain shift as in most previous DG work, we propose a model agnostic training procedure for DG. Our algorithm simulates train/test domain shift during training by synthesizing virtual testing domains within each mini-batch. The meta-optimization objective requires that steps to improve training domain performance should also improve testing domain performance. This meta-learning procedure trains models with good generalization ability to novel domains. We evaluate our method and achieve state of the art results on a recent cross-domain image classification benchmark, as well demonstrating its potential on two classic reinforcement learning tasks.