Do We Really Need to Access the Source Data? Source Hypothesis Transfer for Unsupervised Domain Adaptation
This addresses a practical issue for decentralized private data in machine learning, offering an incremental improvement by enabling adaptation without source data access.
The paper tackles the problem of unsupervised domain adaptation without access to source data by proposing SHOT, a framework that freezes the source model's classifier and learns target-specific features using information maximization and self-supervised pseudo-labeling. Experiments show SHOT achieves state-of-the-art results across various domain adaptation benchmarks.
Unsupervised domain adaptation (UDA) aims to leverage the knowledge learned from a labeled source dataset to solve similar tasks in a new unlabeled domain. Prior UDA methods typically require to access the source data when learning to adapt the model, making them risky and inefficient for decentralized private data. This work tackles a practical setting where only a trained source model is available and investigates how we can effectively utilize such a model without source data to solve UDA problems. We propose a simple yet generic representation learning framework, named \emph{Source HypOthesis Transfer} (SHOT). SHOT freezes the classifier module (hypothesis) of the source model and learns the target-specific feature extraction module by exploiting both information maximization and self-supervised pseudo-labeling to implicitly align representations from the target domains to the source hypothesis. To verify its versatility, we evaluate SHOT in a variety of adaptation cases including closed-set, partial-set, and open-set domain adaptation. Experiments indicate that SHOT yields state-of-the-art results among multiple domain adaptation benchmarks.