Unsupervised Multi-source Domain Adaptation Without Access to Source Data

TL;DR

This paper introduces a novel algorithm for multi-source unsupervised domain adaptation that combines multiple source models without access to source data, achieving performance comparable or superior to the best individual source model.

Contribution

The paper proposes an efficient method to optimally combine multiple source models in UDA without source data, outperforming individual sources and ensuring at least the best source performance.

Findings

The method matches or exceeds the best source model accuracy in most cases.

It outperforms individual source models on several benchmark datasets.

The approach is theoretically justified and empirically validated.

Abstract

Unsupervised Domain Adaptation (UDA) aims to learn a predictor model for an unlabeled domain by transferring knowledge from a separate labeled source domain. However, most of these conventional UDA approaches make the strong assumption of having access to the source data during training, which may not be very practical due to privacy, security and storage concerns. A recent line of work addressed this problem and proposed an algorithm that transfers knowledge to the unlabeled target domain from a single source model without requiring access to the source data. However, for adaptation purposes, if there are multiple trained source models available to choose from, this method has to go through adapting each and every model individually, to check for the best source. Thus, we ask the question: can we find the optimal combination of source models, with no source data and without target labels, whose performance is no worse than the single best source? To answer this, we propose a novel and efficient algorithm which automatically combines the source models with suitable weights in such a way that it performs at least as good as the best source model. We provide intuitive theoretical insights to justify our claim. Furthermore, extensive experiments are conducted on several benchmark datasets to show the effectiveness of our algorithm, where in most cases, our method not only reaches best source accuracy but also outperforms it.