Inferring Latent Domains for Unsupervised Deep Domain Adaptation

TL;DR

This paper presents a deep learning architecture that automatically discovers latent domains within datasets to improve unsupervised domain adaptation, especially when domain labels are unavailable or datasets are mixed.

Contribution

It introduces a novel deep architecture with a domain assignment branch and specialized layers for aligning feature distributions, addressing the challenge of unknown latent domains in UDA.

Findings

Outperforms state-of-the-art domain adaptation methods on benchmarks.

Effectively discovers latent domains without explicit labels.

Improves robustness of classifiers in mixed-domain scenarios.

Abstract

Unsupervised Domain Adaptation (UDA) refers to the problem of learning a model in a target domain where labeled data are not available by leveraging information from annotated data in a source domain. Most deep UDA approaches operate in a single-source, single-target scenario, i.e. they assume that the source and the target samples arise from a single distribution. However, in practice most datasets can be regarded as mixtures of multiple domains. In these cases, exploiting traditional single-source, single-target methods for learning classification models may lead to poor results. Furthermore, it is often difficult to provide the domain labels for all data points, i.e. latent domains should be automatically discovered. This paper introduces a novel deep architecture which addresses the problem of UDA by automatically discovering latent domains in visual datasets and exploiting this information to learn robust target classifiers. Specifically, our architecture is based on two main components, i.e. a side branch that automatically computes the assignment of each sample to its latent domain and novel layers that exploit domain membership information to appropriately align the distribution of the CNN internal feature representations to a reference distribution. We evaluate our approach on publicly available benchmarks, showing that it outperforms state-of-the-art domain adaptation methods.