Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Field-Guide-Inspired Zero-Shot Learning
Utkarsh Mall, Bharath Hariharan, Kavita Bala
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 |
|---|---|---|---|
| replace expert-annotated similar class with a random class (ablation) | ZSL-Interactive · CUB | accuracy: <20% | Hurt |
| use Sibling-variance instead of random attribute acquisition (ablation) | ZSL-Interactive · CUB | accuracy | Improved |
| use Sibling-variance instead of Representation-change (ablation) | ZSL-Interactive · CUB | accuracy | Improved |
| use Representation-change instead of Sibling-variance in later stages (ablation) | ZSL-Interactive · AWA2 | accuracy | Mixed |
| use taxonomy siblings to compute variance instead of variance over all classes (ablation) | Sibling-variance · CUB | accuracy | Improved |
| let learner pick attributes instead of expert picking 10 attributes (ablation) | ZSL-Interactive · CUB (20 novel classes) | accuracy | Improved |
| add one image of the novel class (Image-based query) (ablation) | ZSL-Interactive · AWA2 | accuracy | Improved |
| add one image of the novel class (Image-based query) (ablation) | ZSL-Interactive · CUB, SUN | accuracy | No clear effect |
| use sibling of expert class as the similar class (ablation) | Sibling-variance · CUB | accuracy | Hurt |
| use random similar class instead of sibling (ablation) | Sibling-variance · CUB | accuracy | Hurt |
| add 10% noise to novel attributes (partial, 5 per class) | ZSL-Interactive · partial attributes | accuracy: ~1% drop vs ~5% for traditional ZSL | Hurt |
| use interactive field-guide annotation instead of reduced random vocabulary (ablation) | ZSL-Interactive · CUB, SUN | accuracy | Improved |
Closest papers by meaning
- Make an Omelette with Breaking Eggs: Zero-Shot Learning for Novel Attribute Synthesis (NeurIPS 2022 · similarity 0.91)
- Multi-Cue Zero-Shot Learning With Strong Supervision (CVPR 2016 · similarity 0.90)
- Recovering the Missing Link: Predicting Class-Attribute Associations for Unsupervised Zero-Shot Learning (CVPR 2016 · similarity 0.89)
- Learning Classifiers for Target Domain with Limited or No Labels (ICML 2019 · similarity 0.89)
- Attribute Prototype Network for Zero-Shot Learning (NeurIPS 2020 · similarity 0.88)
- Improved Zero-Shot Classification by Adapting VLMs with Text Descriptions (CVPR 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.
