Unveiling Class-Labeling Structure for Universal Domain Adaptation

TL;DR

This paper introduces a probabilistic approach and a novel adaptation network for Universal Domain Adaptation, effectively identifying shared label sets and improving target domain performance.

Contribution

It proposes a probabilistic method to locate common label sets and a simple universal adaptation network (S-UAN) for better UDA performance.

Findings

S-UAN outperforms state-of-the-art methods in various UDA settings.

The probabilistic approach accurately identifies shared label sets.

Theoretical analysis of generalization bounds enhances understanding of UDA.

Abstract

As a more practical setting for unsupervised domain adaptation, Universal Domain Adaptation (UDA) is recently introduced, where the target label set is unknown. One of the big challenges in UDA is how to determine the common label set shared by source and target domains, as there is simply no labeling available in the target domain. In this paper, we employ a probabilistic approach for locating the common label set, where each source class may come from the common label set with a probability. In particular, we propose a novel approach for evaluating the probability of each source class from the common label set, where this probability is computed by the prediction margin accumulated over the whole target domain. Then, we propose a simple universal adaptation network (S-UAN) by incorporating the probabilistic structure for the common label set. Finally, we analyse the generalization bound focusing on the common label set and explore the properties on the target risk for UDA. Extensive experiments indicate that S-UAN works well in different UDA settings and outperforms the state-of-the-art methods by large margins.