Source-free Domain Adaptation via Avatar Prototype Generation and Adaptation

By Zhen Qiu, Yifan Zhang, Hongbin Lin, Shuaicheng Niu, Yanxia Liu, Qing Du, Mingkui Tan. 2021

212 citations recorded by the source corpus.

We study a practical domain adaptation task, called source-free unsupervised domain adaptation (UDA) problem, in which we cannot access source domain data due to data privacy issues but only a pre-trained source model and unlabeled target data are available. This task, however, is very difficult due to one key challenge: the lack of source data and target domain labels makes model adaptation very challenging. To address this, we propose to mine the hidden knowledge in the source model and exploit it to generate source avatar prototypes (i.e. representative features for each source class) as well as target pseudo labels for domain alignment. To this end, we propose a Contrastive Prototype Generation and Adaptation (CPGA) method. Specifically, CPGA consists of two stages: (1) prototype generation: by exploring the classification boundary information of the source model, we train a prototype generator to generate avatar prototypes via contrastive learning. (2) prototype adaptation: based on the generated source prototypes and target pseudo labels, we develop a new robust contrastive prototype adaptation strategy to align each pseudo-labeled target data to the corresponding source prototypes. Extensive experiments on three UDA benchmark datasets demonstrate the effectiveness and superiority of the proposed method.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Wen Li
2. Yahao Liu
3. Jinhong Deng
4. Lixin Duan
5. Mingkui Tan
6. Tong Chu
7. Shuaicheng Niu
8. Yifan Zhang

In Computer Vision and Pattern Recognition

1. Mingkui Tan
2. Wen Li
3. Yahao Liu
4. Jinhong Deng
5. Lixin Duan
6. Tong Chu
7. Shuaicheng Niu
8. Yifan Zhang

Citations in the corpus

• Towards Stable Test-Time Adaptation in Dynamic Wild World (2023, methodology): “…UDA methods have been proposed to resolve the adaptation problem when source data are absent, such as generative-based methods that generate source images or prototypes from the model (Li et al., 2020; Kundu et al., 2020; Qiu et al., 2021), and information maximization (Liang et al., 2020).”
• Denoised Maximum Classifier Discrepancy for Source-Free Unsupervised Domain Adaptation (2022, methodology): “The experimental results show that the classification accuracy of our method is higher than the current state-of-the-art SFUDA approaches (Qiu et al. 2021; Liang, Hu, and Feng 2020; Li et al. 2020) on the three benchmark datasets.”
• A Comprehensive Survey on Test-Time Adaptation Under Distribution Shifts (2023, unknown): “In contrast, CPGA [108] trains a prototype generator from conditional noises to generate multiple avatar feature prototypes for each class, encouraging that class prototypes are intra-class compact and inter-class separated.”
• Efficient Test-Time Model Adaptation without Forgetting (2022, methodology): “Moreover, we propose an anti-forgetting regularizer to prevent the model parameters Θ changing too much from Θ o . modeling (Li et al., 2020; Kundu et al., 2020; Qiu et al., 2021) or information maximization (Liang et al., 2020).”