Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Distilled Reverse Attention Network for Open-world Compositional Zero-Shot Learning
Yun Li, Zhe Liu, Saurav Jha, Lina Yao
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 non-local and local attention to base model (ablation) | DRANet base model · UT-Zappos (OW-CZSL) | AUC: 25.2 -> 27.3 | Improved |
| add reverse attention and reversal-based classification (ablation) | ANet · UT-Zappos (OW-CZSL) | AUC: 27.3 -> 28.3 | Improved |
| add reverse distillation (mutual non-local/local) (ablation) | RANet · UT-Zappos (OW-CZSL) | AUC: 28.3 -> 28.8 | Improved |
| add reverse distillation (ablation) | RANet · C-GQA (OW-CZSL) | AUC: 0.94 -> 1.05 | Improved |
| use different attention types for attributes (non-local) and objects (local) instead of identical (ablation) | ANet · UT-Zappos (OW-CZSL) | AUC: 27.2 -> 27.3 | Improved |
| swap attention types (local for attributes, non-local for objects) (ablation) | ANet · UT-Zappos (OW-CZSL) | AUC: 27.3 -> 26.3 | Hurt |
| include reversal predictions at inference as (a'+a'_r)*(o'+o'_r) rather than reverse loss only (ablation) | ANet · UT-Zappos (OW-CZSL) | AUC: 27.7 -> 28.3 | Improved |
| use mutual distillation instead of one-sided l-oriented distillation (ablation) | DRANet · UT-Zappos (OW-CZSL) | AUC: 28.1 -> 28.8 | Improved |
| use DRANet versus CompCos on MIT-States | DRANet · MIT-States (OW-CZSL) | HM: 7.9 vs 8.9 | 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.
- KG-SP — KG-SP: Knowledge Guided Simple Primitives for Open World Compositional Zero-Shot Learning (CVPR 2022) · better on 2 of 2 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
- Learning Attention As Disentangler for Compositional Zero-Shot Learning (CVPR 2023 · similarity 0.92)
- Disentangling Before Composing: Learning Invariant Disentangled Features for Compositional Zero-Shot Learning (TPAMI 2024 · similarity 0.91)
- I2CD: An Invertible Causal Framework for Compositional Zero-Shot Learning via Disentangle-Compose-Disentangle (AAAI 2026 · similarity 0.90)
- Context-based and Diversity-driven Specificity in Compositional Zero-Shot Learning (CVPR 2024 · similarity 0.90)
- Beyond Seen Primitive Concepts and Attribute-Object Compositional Learning (CVPR 2024 · similarity 0.90)
- Revealing the Proximate Long-Tail Distribution in Compositional Zero-Shot Learning (AAAI 2024 · similarity 0.87)
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.
