Research map / Detection and segmentation / Open-set and few-shot detection
Learning Attention As Disentangler for Compositional Zero-Shot Learning
Shaozhe Hao, Kai Han, Kwan-Yee K. Wong
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 |
|---|---|---|---|
| Use ADE instead of the strongest listed baseline | CZSL with ViT-B-16 · Clothing16K closed-world | AUC: 90.6 -> 92.4 | Improved |
| Use ADE instead of the strongest listed baseline | CZSL with ViT-B-16 · UT-Zappos50K closed-world | AUC: 34.5 -> 35.1 | Improved |
| Use ADE instead of the strongest listed baseline | CZSL with ViT-B-16 · C-GQA closed-world | AUC: 3.8 -> 5.2 | Improved |
| Use ADE instead of the strongest listed baseline | CZSL with ViT-B-16 · Clothing16K open-world | AUC: 64.1 -> 68.0 | Improved |
| Use ADE instead of the strongest listed baseline | CZSL with ViT-B-16 · UT-Zappos50K open-world | AUC: 25.3 -> 27.1 | Improved |
| Use ADE instead of the strongest listed baseline | CZSL with ViT-B-16 · C-GQA open-world | AUC: 0.94 -> 1.42 | Improved |
| Add composition self-attention (ablation) | ADE on UT-Zappos50K open-world · UT-Zappos50K open-world | AUC: 23.8 -> 25.3 | Improved |
| Add attribute and object self-attention (ablation) | ADE on UT-Zappos50K open-world · UT-Zappos50K open-world | AUC: 25.3 -> 26.7 | Improved |
| Replace attribute and object self-attention with cross-attention (ablation) | ADE on UT-Zappos50K open-world · UT-Zappos50K open-world | AUC: 26.7 -> 26.9 | Improved |
| Add attention-level EMD regularization (ablation) | ADE on UT-Zappos50K open-world · UT-Zappos50K open-world | AUC: 26.9 -> 27.1 | Improved |
| Use attention-level EMD rather than cosine or Euclidean feature distance (ablation) | ADE on UT-Zappos50K open-world · UT-Zappos50K open-world | AUC: 26.8 -> 27.1; 26.2 -> 27.1 | Improved |
| Combine composition probability with attribute-object probability product (ablation) | ADE inference on C-GQA · C-GQA closed-world | AUC: 4.6 -> 5.2 | Improved |
Closest papers by meaning
- Disentangling Before Composing: Learning Invariant Disentangled Features for Compositional Zero-Shot Learning (TPAMI 2024 · similarity 0.92)
- Context-based and Diversity-driven Specificity in Compositional Zero-Shot Learning (CVPR 2024 · similarity 0.89)
- Beyond Seen Primitive Concepts and Attribute-Object Compositional Learning (CVPR 2024 · similarity 0.89)
- A Dynamic Learning Method towards Realistic Compositional Zero-Shot Learning (AAAI 2024 · similarity 0.89)
- Revealing the Proximate Long-Tail Distribution in Compositional Zero-Shot Learning (AAAI 2024 · similarity 0.88)
- LOGICZSL: Exploring Logic-induced Representation for Compositional Zero-shot Learning (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.
