Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Meta Discovery: Learning to Discover Novel Classes given Very Limited Data
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 |
|---|---|---|---|
| add CATA to MM (ablation) | MM · SVHN (5-way), 5- and 1-observation | ACC (%): 40.1±0.3 -> 47.3±0.3; 39.7±0.4 -> 42.7±0.3 | Helped |
| add CATA to MM (ablation) | MM · CIFAR-10 (5-way), 5- and 1-observation | ACC (%): 42.7±0.3 -> 45.3±0.1; 40.3±0.3 -> 41.3±0.4 | Helped |
| add CATA to MM (ablation) | MM · CIFAR-100 (20-way), 5- and 1-observation | ACC (%): 39.9±0.3 -> 42.0±0.4; 37.1±0.3 -> 39.2±0.2 | Helped |
| add CATA to MM (ablation) | MM · OmniGlot (20-way), 5- and 1-observation | ACC (%): 80.1±0.4 -> 82.1±0.4; 78.5±0.4 -> 77.3±0.3 | Mixed |
| add CATA to MP (ablation) | MP · SVHN (5-way), 5- and 1-observation | ACC (%): 60.5±0.2 -> 61.0±0.5; 50.2±0.3 -> 52.8±0.4 | Helped |
| add CATA to MP (ablation) | MP · CIFAR-10 (5-way), 5- and 1-observation | ACC (%): 57.9±0.1 -> 58.5±0.2; 47.6±0.2 -> 51.7±0.2 | Helped |
| add CATA to MP (ablation) | MP · CIFAR-100 (20-way), 5- and 1-observation | ACC (%): 44.1±0.1 -> 45.5±0.3; 37.0±0.2 -> 38.8±0.4 | Helped |
| add CATA to MP (ablation) | MP · OmniGlot (20-way), 5- and 1-observation | ACC (%): 96.7±0.3 -> 98.4±0.2; 91.2±0.2 -> 94.6±0.3 | Helped |
Closest papers by meaning
- When and How Does Known Class Help Discover Unknown Ones? Provable Understanding Through Spectral Analysis (ICML 2023 · similarity 0.90)
- Joint Class-level and Instance-level Relationship Modeling for Novel Class Discovery (AAAI 2025 · similarity 0.89)
- Semantic-Guided Novel Category Discovery (AAAI 2024 · similarity 0.86)
- Adaptive Discovering and Merging for Incremental Novel Class Discovery (AAAI 2024 · similarity 0.86)
- Liberating Seen Classes: Boosting Few-Shot and Zero-Shot Text Classification via Anchor Generation and Classification Reframing (AAAI 2024 · similarity 0.86)
- Generalized Category Discovery via Reciprocal Learning and Class-Wise Distribution Regularization (ICML 2025 · similarity 0.84)
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.