Unsupervised Multi-Target Domain Adaptation: An Information Theoretic Approach

Unsupervised domain adaptation (uDA) models focus on pairwise adaptation settings where there is a single, labeled, source and a single target domain. However, in many real-world settings one seeks to adapt to multiple, but somewhat similar, target domains. Applying pairwise adaptation approaches to this setting may be suboptimal, as they fail to leverage shared information among multiple domains. In this work, we propose an information theoretic approach for domain adaptation in the novel context of multiple target domains with unlabeled instances and one source domain with labeled instances. Our model aims to find a shared latent space common to all domains, while simultaneously accounting for the remaining private, domain-specific factors. Disentanglement of shared and private information is accomplished using a unified information-theoretic approach, which also serves to establish a stronger link between the latent representations and the observed data. The resulting model, accompanied by an efficient optimization algorithm, allows simultaneous adaptation from a single source to multiple target domains. We test our approach on three challenging publicly-available datasets, showing that it outperforms several popular domain adaptation methods.

I. Introduction

In real-world data, the training and test instances often do not come from the same underlying distribution [1]. For example, in the task of object recognition/classification from image data, this may be due to the image noise, changes in the object view, etc., which induce different biases in the observed data sampled during the training and test stage. Consequently, assumptions made by traditional learning algorithms are often violated, resulting in degradation of the algorithms’ performance during inference of test data. Domain Adaptation (DA) approaches [2]–[8] aim to tackle this by transferring knowledge from a source domain (training data) to an unlabeled target domain (test data) to reduce the discrepancy between the source and target data distributions, typically by exploring domain-invariant data structures.