Research map / Label-efficient and robust learning / Domain adaptation and generalization
Prompt Disentanglement via Language Guidance and Representation Alignment for Domain Generalization
De Cheng, Zhipeng Xu, Xinyang Jiang, D. Li, Nannan Wang, Xinbo Gao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add GPT-assisted text disentanglement (GAT) (ablation) | prompt-tuning baseline · Average of 5 DG benchmarks | accuracy: 73.44 -> 74.86 | Improved |
| add image disentanglement guided by text (IMT) (ablation) | baseline · Average of 5 DG benchmarks | accuracy: 73.44 -> 76.69 | Improved |
| add IMT on top of GAT (ablation) | GAT · Average of 5 DG benchmarks | accuracy: 74.86 -> 77.51 | Improved |
| add WERA to GAT+IMT (ablation) | GAT+IMT · Average of 5 DG benchmarks | accuracy: 77.51 -> 78.28 | Improved |
| add DSPL to GAT+IMT (ablation) | GAT+IMT · Average of 5 DG benchmarks | accuracy: 77.51 -> 77.97 | Improved |
| add DSPL to GAT+IMT+WERA (ablation) | GAT+IMT+WERA · Average of 5 DG benchmarks | accuracy: 78.28 -> 78.73 | Improved |
| add WERA to IMT (ablation) | IMT · Average of 5 DG benchmarks | accuracy: 76.69 -> 77.49 | Improved |
| add DSPL to IMT+WERA (ablation) | IMT+WERA · Average of 5 DG benchmarks | accuracy: 77.49 -> 77.81 | Improved |
| add GAT to IMT (ablation) | IMT · TerraInc | accuracy: 55.39 -> 56.82 | Improved |
| add GPT hand-crafted description to baseline text (ablation) | text disentanglement · VLCS/OfficeHome/DomainNet avg | accuracy: 74.73 -> 75.56 | Improved |
| add fine-grained responses with L_a=1 (GPT-3) (ablation) | text disentanglement · VLCS/OfficeHome/DomainNet avg | accuracy: 75.56 -> 75.68 | Improved |
| add domain-invariant responses with L_a=1 (ablation) | text disentanglement · VLCS/OfficeHome/DomainNet avg | accuracy: 75.68 -> 75.66 | No clear effect |
Closest papers by meaning
- Disentangled Prompt Representation for Domain Generalization (CVPR 2024 · similarity 0.96)
- Domain Generalization in CLIP via Learning with Diverse Text Prompts (CVPR 2025 · similarity 0.90)
- A Similarity Paradigm Through Textual Regularization Without Forgetting (AAAI 2025 · similarity 0.90)
- Prompt-Based Distribution Alignment for Unsupervised Domain Adaptation (AAAI 2024 · similarity 0.89)
- Domain-Agnostic Mutual Prompting for Unsupervised Domain Adaptation (CVPR 2024 · similarity 0.89)
- Compound Text-Guided Prompt Tuning via Image-Adaptive Cues (AAAI 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
