Domain Agnostic Learning with Disentangled Representations

TL;DR

This paper introduces a novel deep autoencoder that disentangles domain-specific features from class identity, enabling effective transfer learning across unknown target domains without prior domain labels.

Contribution

The paper proposes the Deep Adversarial Disentangled Autoencoder (DADA), a new method for domain-agnostic learning that does not require prior knowledge of target domain labels.

Findings

DADA achieves state-of-the-art results on multiple image classification datasets.

Disentangling domain-specific features improves transferability to unseen domains.

The approach outperforms existing methods in unsupervised domain transfer scenarios.

Abstract

Unsupervised model transfer has the potential to greatly improve the generalizability of deep models to novel domains. Yet the current literature assumes that the separation of target data into distinct domains is known as a priori. In this paper, we propose the task of Domain-Agnostic Learning (DAL): How to transfer knowledge from a labeled source domain to unlabeled data from arbitrary target domains? To tackle this problem, we devise a novel Deep Adversarial Disentangled Autoencoder (DADA) capable of disentangling domain-specific features from class identity. We demonstrate experimentally that when the target domain labels are unknown, DADA leads to state-of-the-art performance on several image classification datasets.