Domain Generalization with Adversarial Feature Learning

TLDR

The paper addresses domain generalization by learning a feature representation that generalizes to unseen target domains using multiple source domains. The authors propose an adversarial autoencoder framework that aligns domain distributions with Maximum Mean Discrepancy and matches them to a prior via adversarial feature learning, jointly training all components. Experiments on vision tasks show the framework learns more generalizable features than state‑of‑the‑art domain generalization methods.

Abstract

In this paper, we tackle the problem of domain generalization: how to learn a generalized feature representation for an "unseen" target domain by taking the advantage of multiple seen source-domain data. We present a novel framework based on adversarial autoencoders to learn a generalized latent feature representation across domains for domain generalization. To be specific, we extend adversarial autoencoders by imposing the Maximum Mean Discrepancy (MMD) measure to align the distributions among different domains, and matching the aligned distribution to an arbitrary prior distribution via adversarial feature learning. In this way, the learned feature representation is supposed to be universal to the seen source domains because of the MMD regularization, and is expected to generalize well on the target domain because of the introduction of the prior distribution. We proposed an algorithm to jointly train different components of our proposed framework. Extensive experiments on various vision tasks demonstrate that our proposed framework can learn better generalized features for the unseen target domain compared with state-of-the-art domain generalization methods.