Attention-Based Multi-Source Domain Adaptation

Zuo, Yukun; Yao, Hantao; Xu, Changsheng

doi:10.1109/tip.2021.3065254

Cited by 49 publications

(27 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The proposed attention mechanism precisely reduces domain divergence and effectively improves model adaptation towards new target domain.. Some attention-based DA works, such as [38] proposes to use attention for region-level and image-level context learning by exploring relationship in original images and the semantic features; [39] introduces a spatial attention pyramid network to capture context information at different scales; [40] puts forward generative attention adversarial classification network to allows a discriminator to discriminate the transferable regions; [41] proposes attentionbased multi-source DA framework by considering domain correlations and alleviating effect of the dissimilar domains. However, these methods only consider the semantic features in the final model layer, while our work matches the same level of semantic information across model layers through the elaborate dynamic attention mechanism.…”

Section: Attention Mechanismmentioning

confidence: 99%

Attention-based Cross-Layer Domain Alignment for Unsupervised Domain Adaptation

Ma¹,

Yuan²,

Chen³

et al. 2022

Preprint

View full text Add to dashboard Cite

Unsupervised domain adaptation (UDA) aims to learn transferable knowledge from a labeled source domain and adapts a trained model to an unlabeled target domain. To bridge the gap between source and target domains, one prevailing strategy is to minimize the distribution discrepancy by aligning their semantic features extracted by deep models. The existing alignment-based methods mainly focus on reducing domain divergence in the same model layer. However, the same level of semantic information could distribute across model layers due to the domain shifts. To further boost model adaptation performance, we propose a novel method called Attention-based Cross-layer Domain Alignment (ACDA), which captures the semantic relationship between the source and target domains across model layers and calibrates each level of semantic information automatically through a dynamic attention mechanism. An elaborate attention mechanism is designed to reweight each cross-layer pair based on their semantic similarity for precise domain alignment, effectively matching each level of semantic information during model adaptation. Extensive experiments on multiple benchmark datasets consistently show that the proposed method ACDA yields state-of-the-art performance.

show abstract

Section: Attention Mechanismmentioning

confidence: 99%

Attention-based Cross-Layer Domain Alignment for Unsupervised Domain Adaptation

Ma¹,

Yuan²,

Chen³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Increasing works [73], [39], [66], [69], [57] thus focus on multi-source domain adaptation (MSDA) [3] task, where multiple labeled source datasets from different domains are provided for model adaptation. For example, some works [73], [57] present an attention-based strategy to reduce domain divergence in the semantic feature space by using the multiple source datasets and an elaborate attention module.…”

Section: B Domain Adaptationmentioning

confidence: 99%

“…Typical research fields of DA, such as unsupervised domain adaptation (UDA) [60], [68], [29], [28], [34], [67], multisource domain adaptation (MSDA) [73], [39], [66], [69], [57], and multi-target domain adaptation (MTDA) [7], [32], [57], [14], [63], [13] suppose both the source and the target datasets are available for model training. For each new target domain, they have to re-collect target data and use it to repeat the training process, which is expensive, time-consuming, or even infeasible.…”

Section: Introductionmentioning

confidence: 99%

Domain-Specific Bias Filtering for Single Labeled Domain Generalization

Yuan

Chen

et al. 2021

Preprint

View full text Add to dashboard Cite

Domain generalization (DG) utilizes multiple labeled source datasets to train a generalizable model for unseen target domains. However, due to expensive annotation costs, the requirements of labeling all the source data are hard to be met in realworld applications. In this paper, we investigate a Single Labeled Domain Generalization (SLDG) task with only one source domain being labeled, which is more practical and challenging than the Conventional Domain Generalization (CDG). A major obstacle in the SLDG task is the discriminability-generalization bias: discriminative information in the labeled source dataset may contain domain-specific bias, constraining the generalization of the trained model. To tackle this challenging task, we propose a novel method called Domain-Specific Bias Filtering (DSBF), which initializes a discriminative model with the labeled source data and filters out its domain-specific bias with the unlabeled source data for generalization improvement. We divide the filtering process into: (1) Feature extractor debiasing using k-means clustering-based semantic feature re-extraction; and (2) Classifier calibrating using attention-guided semantic feature projection. DSBF unifies the exploration of the labeled and the unlabeled source data to enhance the discriminability and generalization of the trained model, resulting in a highly generalizable model. We further provide theoretical analysis to verify the proposed domain-specific bias filtering process. Extensive experiments on multiple datasets show the superior performance of DSBF in tackling both the challenging SLDG task and the CDG task.

show abstract

“…Unsupervised domain adaptation (UDA) [15], [37], [22], [31], [32], [73] aims to transfer the knowledge learned from the labeled source domain to the unlabeled target domain. It has been widely applied in classification [38], detection [64], and segmentation [71].…”

Section: Introductionmentioning

confidence: 99%

Reciprocal Normalization for Domain Adaptation

Huang¹,

Sheng²,

Li³

et al. 2021

Preprint

View full text Add to dashboard Cite

Batch normalization (BN) is widely used in modern deep neural networks, which has been shown to represent the domain-related knowledge, and thus is ineffective for cross-domain tasks like unsupervised domain adaptation (UDA). Existing BN variant methods aggregate source and target domain knowledge in the same channel in normalization module. However, the misalignment between the features of corresponding channels across domains often leads to a suboptimal transferability. In this paper, we exploit the cross-domain relation and propose a novel normalization method, Reciprocal Normalization (RN). Specifically, RN first presents a Reciprocal Compensation (RC) module to acquire the compensatory for each channel in both domains based on the cross-domain channel-wise correlation. Then RN develops a Reciprocal Aggregation (RA) module to adaptively aggregate the feature with its cross-domain compensatory components. As an alternative to BN, RN is more suitable for UDA problems and can be easily integrated into popular domain adaptation methods. Experiments show that the proposed RN outperforms existing normalization counterparts by a large margin and helps state-of-the-art adaptation approaches achieve better results. The source code is available on https: //github.com/Openning07/reciprocal-normalization-for-DA.

show abstract

Attention-Based Multi-Source Domain Adaptation

Cited by 49 publications

References 20 publications

Attention-based Cross-Layer Domain Alignment for Unsupervised Domain Adaptation

Attention-based Cross-Layer Domain Alignment for Unsupervised Domain Adaptation

Domain-Specific Bias Filtering for Single Labeled Domain Generalization

Reciprocal Normalization for Domain Adaptation

Contact Info

Product

Resources

About