Domain Adaptation without Model Transferring

TL;DR

This paper introduces a novel domain adaptation method that does not require transferring source models, enhancing data privacy and security while maintaining effective adaptation.

Contribution

It proposes a new approach for domain adaptation that refines source model information without transferring models, using Distributionally Adversarial Training for better data alignment.

Findings

Effective on multiple benchmark datasets

Avoids privacy risks associated with model transfer

Achieves promising adaptation performance

Abstract

In recent years, researchers have been paying increasing attention to the threats brought by deep learning models to data security and privacy, especially in the field of domain adaptation. Existing unsupervised domain adaptation (UDA) methods can achieve promising performance without transferring data from source domain to target domain. However, UDA with representation alignment or self-supervised pseudo-labeling relies on the transferred source models. In many data-critical scenarios, methods based on model transferring may suffer from membership inference attacks and expose private data. In this paper, we aim to overcome a challenging new setting where the source models cannot be transferred to the target domain. We propose Domain Adaptation without Source Model, which refines information from source model. In order to gain more informative results, we further propose Distributionally Adversarial Training (DAT) to align the distribution of source data with that of target data. Experimental results on benchmarks of Digit-Five, Office-Caltech, Office-31, Office-Home, and DomainNet demonstrate the feasibility of our method without model transferring.