Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Label-Embedding for Image Classification
Zeynep Akata, Florent Perronnin, Zaïd Harchaoui, Cordelia Schmid
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 continuous attribute embeddings instead of binary {-1,+1} (ablation) | ALE · AWA zero-shot (64K FV) | top-1 accuracy (%): 41.8 -> 48.5 | Improved |
| use continuous attribute embeddings instead of binary {-1,+1} (ablation) | ALE · CUB zero-shot (64K FV) | top-1 accuracy (%): 19.6 -> 26.9 | Improved |
| add mean-centering of output embeddings (ablation) | ALE · AWA zero-shot (64K FV) | top-1 accuracy (%): 48.5 -> 47.7 | No clear effect |
| add l2-normalization of output embeddings (ablation) | ALE · AWA zero-shot (64K FV, continuous) | top-1 accuracy (%): 44.9 -> 48.5 | Improved |
| add l2-normalization of output embeddings (ablation) | ALE · CUB zero-shot (64K FV, continuous) | top-1 accuracy (%): 22.7 -> 26.9 | Improved |
| use ranking objective instead of ridge regression (ablation) | ALE · AWA zero-shot | top-1 accuracy (%): 44.5 -> 48.5 | Improved |
| use ranking objective instead of ridge regression (ablation) | ALE · CUB zero-shot | top-1 accuracy (%): 21.6 -> 26.3 | Improved |
| use ALE instead of DAP | DAP · CUB zero-shot | top-1 accuracy (%): 12.3 -> 26.9 | Improved |
| use ALE instead of DAP | DAP · AWA zero-shot | top-1 accuracy (%): 41.0 -> 48.5 | Improved |
| use ALE instead of DAP for attribute prediction | DAP · CUB zero-shot | attribute AUC (%): 64.8 -> 59.4 | Hurt |
| replace attribute embeddings with Word2Vec (WLE) (ablation) | ALE · AWA zero-shot | top-1 accuracy (%): 48.5 -> 32.5 | Hurt |
| combine attributes and hierarchies with late fusion (AHLE late) | ALE · AWA zero-shot | top-1 accuracy (%): 48.5 -> 49.4 | Improved |
Closest papers by meaning
- Recovering the Missing Link: Predicting Class-Attribute Associations for Unsupervised Zero-Shot Learning (CVPR 2016 · similarity 0.89)
- Disentangling Visual Embeddings for Attributes and Objects (CVPR 2022 · similarity 0.88)
- Semantic Diversity Learning for Zero-Shot Multi-Label Classification (ICCV 2021 · similarity 0.88)
- Attribute Prototype Network for Any-Shot Learning (IJCV 2022 · similarity 0.87)
- Attributes2Classname: A Discriminative Model for Attribute-Based Unsupervised Zero-Shot Learning (ICCV 2017 · similarity 0.87)
- Navigating Real-World Partial Label Learning: Unveiling Fine-Grained Images with Attributes (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.
