Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Few-Shot One-Class Classification via Meta-Learning
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 |
|---|---|---|---|
| Use one-class adaptation batches and class-balanced validation batches for meta-training MAML | MAML · MiniImageNet, Omniglot, MT-MNIST T0, and STS-Sawtooth | accuracy: 62.3 -> 69.1; 91.4 -> 96.6; 85.5 -> 88; 81.1 -> 96.6 (K=2) | Helped |
| Use one-class episode sampling with MetaOptNet and an OC-SVM inner classifier | MetaOptNet · MiniImageNet, CIFAR-FS, and FC100 | accuracy: 50 -> 51.8; 56 -> 56.3; 51.2 -> 52.2 (K=2); 56.6 -> 67.4; 74.8 -> 75.5; 53.3 -> 59.9 (K=10) | Helped |
| Use one-class episode sampling with Meta-SGD | MetaSGD · MiniImageNet, CIFAR-FS, and FC100 | accuracy: 65 -> 69.6; 58.4 -> 71.4; 55 -> 60.3 (K=2); 73.6 -> 75.8; 71.3 -> 77.8; 61.3 -> 64.3 (K=10) | Helped |
| Adapt first-order meta-learning algorithms using one-class episodes | FOMAML and Reptile · MiniImageNet, Omniglot, MT-MNIST T0, and STS-Sawtooth | accuracy: 53.3 -> 55.7; 78.8 -> 74.7; 80.7 -> 79.1; 75.1 -> 58.6 (OC-FOMAML vs FOMAML, K=2) | Hurt |
| Use batch normalization (ablation) | OC-MAML · Image datasets | accuracy: 66.4 -> 69.1 (MiniImageNet, K=2); 95.6 -> 96.6 (Omniglot, K=2); 85.2 -> 88 (MT-MNIST, K=2) | Helped |
| Use batch normalization (ablation) | OC-MAML · STS-Sawtooth, K=2 | accuracy: 96.6 -> 51.3 | Hurt |
| Use one-class episode sampling instead of standard class-balanced sampling | MAML · MiniImageNet, CIFAR-FS, and FC100 | accuracy: 62.3 -> 69.1; 62.1 -> 70; 55.1 -> 59.9 (K=2); 65.5 -> 76.2; 69.1 -> 79.1; 61.6 -> 65.5 (K=10) | Helped |
| Use OC-MAML for CNC-MMD adaptation with K=10 normal examples | OC-MAML · CNC-MMD finishing and roughing operations | F1-score | Helped |
Closest papers by meaning
- Task-Specific Gradient Adaptation for Few-Shot One-Class Classification (CVPR 2025 · similarity 0.89)
- A Closer Look at Few-shot Classification Again (ICML 2023 · similarity 0.89)
- How to Train Your MAML to Excel in Few-Shot Classification (ICLR 2022 · similarity 0.88)
- Meta-Dataset: A Dataset of Datasets for Learning to Learn from Few Examples (ICLR 2020 · similarity 0.88)
- MetaCoCo: A New Few-Shot Classification Benchmark with Spurious Correlation (ICLR 2024 · similarity 0.88)
- Unraveling Meta-Learning: Understanding Feature Representations for Few-Shot Tasks (ICML 2020 · similarity 0.88)
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.