Research map / Label-efficient and robust learning / Domain adaptation and generalization
Learning Attributes Equals Multi-Source Domain Generalization
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 category-invariant UDICA features instead of the original features | Attribute detectors · AWA and CUB | AUC: 72.8/{78.9}^{*} -> 83.9; 61.8/{72.1 }^{*} -> 76.0 | Helped |
| Add attribute-label kernel alignment to UDICA (ablation) | UDICA attribute representations · AWA, CUB, a-Yahoo, and UCF101 | AUC: 83.9 -> 84.4; 76.0 -> 76.4; 82.3 -> 84.7; 74.3 ± 1.3 -> 75.5 ± 1.1 | Helped |
| Use UDICA attribute representations in the DAP zero-shot pipeline | DAP · AWA, CUB, and UCF101 | accuracy: 49.4* -> 63.6; 34.9* -> 42.4; 26.8 ± 1.1 -> 29.6 ± 1.2 | Helped |
| Add attribute-label kernel alignment to UDICA features in the DAP zero-shot pipeline (ablation) | UDICA + DAP · AWA, CUB, and UCF101 | accuracy: 63.6 -> 73.8; 42.4 -> 43.7; 29.6 ± 1.2 -> 31.1 ± 0.8 | Helped |
| Use UDICA features instead of VGG features with late fusion | Multi-attribute retrieval · AWA single-, double-, and triple-attribute queries | AUC: 78.9 -> 83.9; 77.2 -> 79.5; 76.1 -> 78.6 | Helped |
| Use KDICA features instead of VGG features with late fusion | Multi-attribute retrieval · AWA single-, double-, and triple-attribute queries | AUC: 78.9 -> 84.4; 77.2 -> 81.0; 76.1 -> 79.4 | Helped |
| Use UDICA features instead of VGG features with TagProp | Multi-attribute retrieval · AWA single-, double-, and triple-attribute queries | AUC: 76.3 -> 78.5; 75.9 -> 76.1; 75.5 -> 75.6 | Mixed |
| Use KDICA features instead of VGG features with TagProp | Multi-attribute retrieval · AWA single-, double-, and triple-attribute queries | AUC: 76.3 -> 79.2; 75.9 -> 76.1; 75.5 -> 75.8 | Mixed |
| Add centered kernel alignment to the UDICA representation (ablation) | Attribute detectors · Attributes shared across visually different categories | attribute prediction performance | Helped |
| Apply category-invariant feature learning to attributes already near 100% detectable | Attribute detectors · Attributes such as yellow | attribute prediction performance | No clear effect |
Closest papers by meaning
- DoGA: Enhancing Grounded Object Detection via Grouped Pre-Training with Attributes (AAAI 2025 · similarity 0.87)
- Automated Detection of Visual Attribute Reliance with a Self-Reflective Agent (NeurIPS 2025 · similarity 0.87)
- Compositional Caching for Training-free Open-vocabulary Attribute Detection (CVPR 2025 · similarity 0.87)
- Detecting Open World Objects via Partial Attribute Assignment (CVPR 2025 · similarity 0.86)
- Black Sheep in the Herd: Playing with Spuriously Correlated Attributes for Vision-Language Recognition (ICLR 2025 · similarity 0.86)
- Improving Single Domain-Generalized Object Detection: A Focus on Diversification and Alignment (CVPR 2024 · similarity 0.86)
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.