Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Learning Attributes from the Crowdsourced Relative Labels
Read the paperCheck 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 85 AwA attributes instead of crowdsourced attributes (K=15, rho=0.60 compared) | ESZSL with VGG19 · AwA zero-shot, MT=2 | accuracy: 0.762 ± 0.184 -> 0.794 ± 0.162 | Helped |
| crowdsourced attributes versus complete AwA-85 attributes | ESZSL with VGG19 · AwA zero-shot, MT=5 | accuracy: 0.453 ± 0.128 -> 0.502 ± 0.128 | Helped |
| crowdsourced attributes versus best random AwA subset of equal size (2K entries) | ESZSL with VGG19 · AwA zero-shot, MT=2, rho=0.60 | accuracy: 0.763 ± 0.204 -> 0.794 ± 0.162 | Helped |
| crowdsourced attributes versus best random AwA subset of equal size | ESZSL with VGG19 · AwA zero-shot, MT=3, rho=0.70 | accuracy: 0.626 ± 0.170 -> 0.673 ± 0.160 | Helped |
| crowdsourced attributes versus best random AwA subset of equal size | ESZSL with VGG19 · AwA zero-shot, MT=5, rho=0.55 | accuracy: 0.439 ± 0.120 -> 0.442 ± 0.149 | No clear effect |
| increase rho from 0.55 to 0.70 (K from 11 to 28) | crowdsourced attributes · AwA zero-shot, MT=2 | accuracy: 0.793 ± 0.189 -> 0.822 ± 0.144 | Helped |
| increase rho from 0.55 to 0.70 (K from 11 to 28) | crowdsourced attributes · AwA zero-shot, MT=4 | accuracy: 0.572 ± 0.186 -> 0.592 ± 0.146 | Mixed |
| increase rho from 0.65 to 0.70 (K from 20 to 28) | crowdsourced attributes · AwA zero-shot, MT=5 | accuracy: 0.502 ± 0.128 -> 0.478 ± 0.124 | Hurt |
| increase rho from 0.55 to 0.70 (K from 11 to 28) | AwA-Best 2K entries · AwA zero-shot, MT=3 | accuracy: 0.604 ± 0.169 -> 0.626 ± 0.170 | Mixed |
Closest papers by meaning
- Towards a Foundation Model for Crowdsourced Label Aggregation (ICLR 2026 · similarity 0.86)
- From Common to Special: When Multi-Attribute Learning Meets Personalized Opinions (AAAI 2018 · similarity 0.86)
- Towards Professional Level Crowd Annotation of Expert Domain Data (CVPR 2023 · similarity 0.86)
- Unbiased Multi-Label Learning from Crowdsourced Annotations (ICML 2024 · similarity 0.84)
- Efficient Online Crowdsourcing with Complex Annotations (AAAI 2024 · similarity 0.83)
- Bayesian Inference for Correlated Human Experts and Classifiers (ICML 2025 · similarity 0.83)
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.