Model Adaptation: Unsupervised Domain Adaptation Without Source Data

By Rui Li, Qianfen Jiao, Wenming Cao, Hau-San Wong, Si Wu. 2020

585 citations recorded by the source corpus.

In this paper, we investigate a challenging unsupervised domain adaptation setting --- unsupervised model adaptation. We aim to explore how to rely only on unlabeled target data to improve performance of an existing source prediction model on the target domain, since labeled source data may not be available in some real-world scenarios due to data privacy issues. For this purpose, we propose a new framework, which is referred to as collaborative class conditional generative adversarial net to bypass the dependence on the source data. Specifically, the prediction model is to be improved through generated target-style data, which provides more accurate guidance for the generator. As a result, the generator and the prediction model can collaborate with each other without source data. Furthermore, due to the lack of supervision from source data, we propose a weight constraint that encourages similarity to the source model. A clustering-based regularization is also introduced to produce more discriminative features in the target domain. Compared to conventional domain adaptation methods, our model achieves superior performance on multiple adaptation tasks with only unlabeled target data, which verifies its effectiveness in this challenging setting.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning

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

In Computer Vision and Pattern Recognition

In Representation & Self-Supervised Learning

Citations in the corpus

Refining Language Models with Compositional Explanations (2021, background): “, due to privacy constraints [27], or access restriction [29]) but unlabeled data in the target domain (DT ) are readily accessible, during model refinement.”
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).”
Source-Free Domain Adaptation for Semantic Segmentation (2021, methodology): “Recently, a tiny number of source-free UDA methods [25, 24, 27, 38, 22, 26] have been developed to tackle a similar issue on image classification.”
A Comprehensive Survey on Test-Time Adaptation Under Distribution Shifts (2023, unknown): “(7), another line of SFDA methods [12], [127], [242] attempts to encourage consistency between target samples with their data-level neighbors, and APA [130] learns the neighbors in the feature space.”
Denoised Maximum Classifier Discrepancy for Source-Free Unsupervised Domain Adaptation (2022, unknown): “…76.4 STAR (Lu et al. 2020) 95.0 84.0 84.6 73.0 91.6 91.8 85.9 78.4 94.4 84.7 87.0 42.2 82.7 BCDM (Li et al. 2021) 95.1 87.6 81.2 73.2 92.7 95.4 86.9 82.5 95.1 84.8 88.1 39.5 83.4 SHOT (Liang, Hu, and Feng 2020) 94.3 88.5 80.1 57.3 93.1 94.9 80.7 80.3 91.5 89.1 86.3 58.2 82.9 MA (Li et al. 2020) 94.”
Test-Time Classifier Adjustment Module for Model-Agnostic Domain Generalization (2021, background): “Among them, recent “source-free” setups are particularly similar to our setting [30, 25, 34].”
Contrastive Test-Time Adaptation (2022, methodology): “To address these difficulties, existing works have leveraged image/feature generation [23, 27], class prototypes [28, 57], entropy minimization [28, 56], self-training or pseudo labeling [28], and self-supervised auxiliary task training [50].”
FixBi: Bridging Domain Spaces for Unsupervised Domain Adaptation (2020, unknown): “5 MODEL [19] 94.”
Exploiting the Intrinsic Neighborhood Structure for Source-free Domain Adaptation (2021, methodology): “SHOT [21] and 3C-GAN [20] are for closed-set DA where source and target domains have the same categories.”
Overcoming Negative Transfer: A Survey (2020, background): “Some existing works actually follow this protocol [103], [104].”
Tent: Fully Test-Time Adaptation by Entropy Minimization (2021, methodology): “Li et al. (2020); Kundu et al. (2020) rely on generative modeling and optimize multiple models with multiple losses.”
Continual Test-Time Domain Adaptation (2022, background): “Some existing works [29,33,68] utilize generative models to support the feature alignment in absence of source data.”
Give Me Your Trained Model: Domain Adaptive Semantic Segmentation without Source Data (2021, methodology): “To tackle source data-free domain adaptation problem, a few recent methods [28–30] provide kinds of model adaptation solutions for classiﬁcation problems.”
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).”