Jacobian Norm for Unsupervised Source-Free Domain Adaptation

TL;DR

This paper introduces a Jacobian norm regularizer guided by model smoothness to improve unsupervised source-free domain adaptation, backed by theoretical error bounds and extensive experiments showing superior performance.

Contribution

It provides the first theoretical generalization error bound based on model smoothness for USFDA and proposes a simple Jacobian norm regularizer to enhance adaptation performance.

Findings

The Jacobian norm regularizer improves target domain generalization.

Theoretical error bounds support the effectiveness of the approach.

Experimental results outperform existing USFDA methods.

Abstract

Unsupervised Source (data) Free domain adaptation (USFDA) aims to transfer knowledge from a well-trained source model to a related but unlabeled target domain. In such a scenario, all conventional adaptation methods that require source data fail. To combat this challenge, existing USFDAs turn to transfer knowledge by aligning the target feature to the latent distribution hidden in the source model. However, such information is naturally limited. Thus, the alignment in such a scenario is not only difficult but also insufficient, which degrades the target generalization performance. To relieve this dilemma in current USFDAs, we are motivated to explore a new perspective to boost their performance. For this purpose and gaining necessary insight, we look back upon the origin of the domain adaptation and first theoretically derive a new-brand target generalization error bound based on the model smoothness. Then, following the theoretical insight, a general and model-smoothness-guided Jacobian norm (JN) regularizer is designed and imposed on the target domain to mitigate this dilemma. Extensive experiments are conducted to validate its effectiveness. In its implementation, just with a few lines of codes added to the existing USFDAs, we achieve superior results on various benchmark datasets.