[IEEE TPAMI 2024] GH/GH++: Gradient Harmonization in Unsupervised Domain Adaptation The paper "Gradient Harmonization in Unsupervised Domain Adaptation" has been accepted by IEEE TPAMI 2024. Contents

2024年8月1日 · In this paper, we delve into this issue and introduce two effective solutions known as Gradient Harmonization, including GH and GH++, to mitigate the conflict between domain alignment and classification tasks.

2024年8月5日 · In this paper, we delve into this issue and introduce two effective solutions known as Gradient Harmonization, including GH and GH++, to mitigate the conflict between domain alignment and classification tasks.

[IEEE TPAMI 2024] GH/GH++: Gradient Harmonization in Unsupervised Domain Adaptation

issue and introduce two effective solutions known as Gradient Harmonization, including GH and GH++, to mitigate the conflict between domain alignment and classification tasks. GH operates by altering the gradient angle between different tasks from an obtuse angle to

The proposed GH/GH++ is a universal plug-and-play approach for between-task balance learning and can be easily embedded in most alignment-based unsupervised domain adaptation methods. This work is a new upgrade and more general form of our previous CVPR paper from the perspective of gradient harmonization.

To alleviate these issues, we propose a Domain-guided Conditional Diffusion Model (DCDM), which generates high-fidelity target domain samples, making the transfer from source domain to target domain easier.

2024年9月9日 · In this paper, we delve into this issue and introduce two effective solutions known as Gradient Harmonization, including GH and GH++, to mitigate the conflict between domain alignment and classification tasks.

Domain adaptation attempts to boost the performance on a target domain by borrowing knowledge from a well established source domain. To handle the distribution.

2024年3月24日 · We provide a comprehensive theoretical analysis of UGDA and derive a generalization bound for multi-layer GNNs. By formulating GNN Lipschitz for k-layer GNNs, we show that the target risk bound can be tighter by removing propagation layers in source graph and stacking multiple propagation layers in target graph.