Attracting and Dispersing: A Simple Approach for Source-free Domain Adaptation

NeurIPS 2022 Spotlight

Shiqi Yang†, Yaxing Wang‡ ,Kai Wang†, Shangling Jui *, Joost van de Weijer

† Computer Vision Center UAB, Barcelona, Spain

Nankai University, China

* Huawei Kirin Solution, Shanghai, China


We propose a simple but effective source-free domain adaptation (SFDA) method. Treating SFDA as an unsupervised clustering problem and following the intuition that local neighbors in feature space should have more similar predictions than other features, we propose to optimize an objective of prediction consistency. This objective encourages local neighborhood features in feature space to have similar predictions while features farther away in feature space have dissimilar predictions, leading to efficient feature clustering and cluster assignment simultaneously. For efficient training, we seek to optimize an upper-bound of the objective resulting in two simple terms. Furthermore, we relate popular existing methods in domain adaptation, source-free domain adaptation and contrastive learning via the perspective of discriminability and diversity. The experimental results prove the superiority of our method, and our method can be adopted as a simple but strong baseline for future research in SFDA. Our method can be also adapted to source-free open-set and partial-set DA which further shows the generalization ability of our method.


We achieve adaptation by optimizing an upper-bound of the clustering objectively:

MI: Mutual Information (SHOT ICML 2020)

BNM: Batch Nuclear-norm Maximization (CVPR 2020)

NC: Neighborhood Clustering (ICCV 2021 and NeurIPS 2021)