Research map / Label-efficient and robust learning / Domain adaptation and generalization
Domain Generalization via Text-Anchored Information Bottleneck
Eunyi Lyou, YUNJEONG CHOI, Junho Lee, Lee Joonseok
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 L_comp compression loss (ablation) | L_sem + L_align · OfficeHome (ResNet-50) | accuracy: 76.0 -> 79.0 | Improved |
| add L_align alignment loss (ablation) | L_sem + L_comp · OfficeHome (ResNet-50) | accuracy: 78.9 -> 79.0 | Improved |
| add L_comp compression loss (ablation) | L_sem · DomainNet (ResNet-50) | accuracy: 35.5 -> 57.6 | Improved |
| add L_align alignment loss (ablation) | L_sem · DomainNet (ResNet-50) | accuracy: 35.5 -> 37.3 | Improved |
| add L_align alignment loss (ablation) | L_sem + L_comp · OfficeHome (ViT) | accuracy: 85.1 -> 86.4 | Improved |
| add L_align alignment loss (ablation) | L_sem · PACS (CLIP) | accuracy: 96.8 -> 98.2 | Improved |
| replace random anchors with CLIP text anchors (ablation) | Ours · OfficeHome (ResNet-50) | accuracy: 76.6 -> 79.0 | Improved |
| replace random anchors with CLIP text anchors (ablation) | Ours · DomainNet (ResNet-50) | accuracy: 36.4 -> 58.3 | Improved |
| use MiniLM language encoder instead of CLIP text encoder (ablation) | Ours · OfficeHome | accuracy: 79.0 -> 79.1 | No clear effect |
| use ImageNet prompts instead of 'a photo of [cls]' with CLIP (ablation) | Ours · OfficeHome | accuracy: 79.0 -> 77.9 | Hurt |
| fix anchors (freeze) with the proposed objective (ablation) | Ours · OfficeHome (CLIP anchors) | accuracy: 77.2 -> 79.0 | Improved |
| fix anchors with the L2 objective (ablation) | L2 alignment · OfficeHome (random anchors) | accuracy: 13.4 -> 11.9 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CLIPood — CLIPood: Generalizing CLIP to Out-of-Distributions (ICML 2023) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Leveraging Vision-Language Models for Improving Domain Generalization in Image Classification (CVPR 2024 · similarity 0.89)
- Explaining Domain Shifts in Language: Concept Erasing for Interpretable Image Classification (CVPR 2025 · similarity 0.89)
- PracticalDG: Perturbation Distillation on Vision-Language Models for Hybrid Domain Generalization (CVPR 2024 · similarity 0.88)
- DGFamba: Learning Flow Factorized State Space for Visual Domain Generalization (AAAI 2025 · similarity 0.88)
- Disentangled Prompt Representation for Domain Generalization (CVPR 2024 · similarity 0.88)
- Domain Generalization in CLIP via Learning with Diverse Text Prompts (CVPR 2025 · similarity 0.88)
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.
