Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Discriminative Learning of Latent Features for Zero-Shot Recognition
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 |
|---|---|---|---|
| Fine-tune the image feature network end-to-end instead of keeping image features fixed (ablation) | Single-scale baseline embedding (SS-BE) · AwA and CUB | MCA (%): 75.20 -> 79.35; 50.51 -> 59.32 | Helped |
| Add latent attributes alongside user-defined attributes (ablation) | SS-BE-Learned · AwA and CUB | MCA (%): 79.35 -> 81.36; 59.32 -> 65.99 | Helped |
| Add zoom-based coarse-to-fine scales (ablation) | SS-BE-Learned · AwA and CUB | MCA (%): 79.35 -> 81.80; 59.32 -> 64.85 | Helped |
| Combine user-defined and latent attribute features for prediction (ablation) | SS-AE-Learned · AwA and CUB | MCA (%): 80.97 -> 81.36; 62.17 -> 65.99 | Helped |
| Jointly train region mining and latent-attribute modeling (ablation) | LDF · AwA and CUB | MCA (%): 78.76 -> 79.35; 63.08 -> 66.47 | Helped |
| Jointly train region mining and latent-attribute modeling (ablation) | LDF · AwA and CUB | MCA (%): 81.80 -> 82.47; 64.85 -> 65.94 | Helped |
| Use both user-defined and latent features at prediction (ablation) | LDF · AwA and CUB | MCA (%): 82.47 -> 83.40; 65.94 -> 67.12 | Helped |
| Add a third zoom scale (ablation) | MS-BE-Learned on AwA · AwA | MCA (%): 77.12 -> 77.05 | No clear effect |
| Increase latent-attribute dimension from k to 2k or 3k (ablation) | SS-AE-Learned (LA) · AwA | MCA (%): 75.75 -> 75.83 -> 76.01 | No clear effect |
| Replace the learned zoom region with 2,000 EdgeBox proposals and MIL selection (ablation) | SS-BE baseline · AwA | MCA (%): 78.35 -> 72.67% | Hurt |
| Increase from scale 1 to scale 2 (ablation) | MS-BE-Learned · AwA and CUB | MCA (%): 79.20 -> 79.87; 59.88 -> 61.04 | Helped |
| Use latent features alone instead of user-defined features alone (ablation) | SS-AE-Learned · AwA and CUB | MCA (%): 80.97 -> 78.76; 62.17 -> 63.08 | Mixed |
Closest papers by meaning
- Attentive Region Embedding Network for Zero-Shot Learning (CVPR 2019 · similarity 0.92)
- Learning a Deep Embedding Model for Zero-Shot Learning (CVPR 2017 · similarity 0.91)
- Context-Aware Zero-Shot Learning for Object Recognition (ICML 2019 · similarity 0.91)
- ZeroDiff: Solidified Visual-semantic Correlation in Zero-Shot Learning (ICLR 2025 · similarity 0.90)
- Data-Free Generalized Zero-Shot Learning (AAAI 2024 · similarity 0.90)
- ZeroMamba: Exploring Visual State Space Model for Zero-Shot Learning (AAAI 2025 · similarity 0.90)
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.