Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Predicting Visual Exemplars of Unseen Classes for Zero-Shot Learning
Soravit Changpinyo, Wei-Lun Chao, Fei Sha
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 |
|---|---|---|---|
| predict visual exemplars and use as semantic representations instead of raw attributes | SynC struct · AwA | accuracy (%): 72.9 -> 77.2 | Improved |
| use predicted exemplars with ConSE | ConSE · CUB | accuracy (%): 36.2 -> 46.2 | Improved |
| use predicted exemplars with LatEm | LatEm · CUB | accuracy (%): 48.0 -> 56.2 | Improved |
| use predicted exemplars with ConSE | ConSE · AwA | accuracy (%): 63.3 -> 70.5 | Improved |
| use predicted exemplars with SynC struct | SynC struct · SUN | accuracy (%): 62.7 -> 66.1 | Improved |
| use 1NN on predicted exemplars instead of SynC o-vs-o | SynC o-vs-o · ImageNet 2-hop | Flat Hit@1 (%): 10.5 -> 11.7 | Improved |
| use standardized Euclidean distance (1NNs) instead of Euclidean (1NN) (ablation) | EXEM (1NN) · CUB | accuracy (%): 56.3 -> 58.5 | Improved |
| use standardized Euclidean distance (1NNs) instead of Euclidean (1NN) (ablation) | EXEM (1NN) · SUN | accuracy (%): 69.6 -> 67.3 | Hurt |
| add PCA projection d=1024 before regression (ablation) | EXEM (1NN) · AwA | accuracy (%): 77.8 -> 76.2 | Hurt |
| add PCA projection d=1024 before regression (ablation) | EXEM (1NN) · CUB | accuracy (%): 55.1 -> 56.3 | Improved |
| reduce PCA dimension from 1024 to 500 (ablation) | EXEM (1NN) · AwA | accuracy (%): 76.2 -> 76.2 | No clear effect |
| use SVR instead of 2-layer MLP for exemplar prediction (ablation) | EXEM (1NN) · CUB split 1 | accuracy (%): 53.8 -> 57.1 | Improved |
Closest papers by meaning
- Classifier and Exemplar Synthesis for Zero-Shot Learning (IJCV 2019 · similarity 0.92)
- Synthesized Classifiers for Zero-Shot Learning (CVPR 2016 · similarity 0.92)
- Rethinking Zero-Shot Learning: A Conditional Visual Classification Perspective (ICCV 2019 · similarity 0.92)
- SimZSL: Zero-Shot Learning Beyond a Pre-defined Semantic Embedding Space (IJCV 2025 · similarity 0.92)
- VGSE: Visually-Grounded Semantic Embeddings for Zero-Shot Learning (CVPR 2022 · similarity 0.92)
- Progressive Semantic-Guided Vision Transformer for Zero-Shot Learning (CVPR 2024 · 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.
