A Theory of Output-Side Unsupervised Domain Adaptation

TL;DR

This paper develops a theoretical framework for output-side unsupervised domain adaptation, analyzing scenarios where only target outputs or both inputs and outputs are available post-mapping, with bounds based on discrepancy measures.

Contribution

It introduces a theoretical analysis of output-side domain adaptation, including new bounds for various data availability scenarios, extending prior input-focused approaches.

Findings

Derived generalization bounds using discrepancy measures.

Analyzed three variants of output-side domain adaptation.

Provided theoretical insights into domain transfer problems.

Abstract

When learning a mapping from an input space to an output space, the assumption that the sample distribution of the training data is the same as that of the test data is often violated. Unsupervised domain shift methods adapt the learned function in order to correct for this shift. Previous work has focused on utilizing unlabeled samples from the target distribution. We consider the complementary problem in which the unlabeled samples are given post mapping, i.e., we are given the outputs of the mapping of unknown samples from the shifted domain. Two other variants are also studied: the two sided version, in which unlabeled samples are give from both the input and the output spaces, and the Domain Transfer problem, which was recently formalized. In all cases, we derive generalization bounds that employ discrepancy terms.