Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Improved Few-Shot Visual Classification
Check 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 |
|---|---|---|---|
| Replace CNAPS's adapted linear classifier with class-covariance Mahalanobis classification | CNAPS · Meta-Dataset overall average | accuracy: 65.9±0.8 -> 72.2±0.8 | Helped |
| Use autoregressive feature adaptation (ablation) | Simple CNAPS · Meta-Dataset overall average | accuracy: 72.2±0.8 -> 71.2±0.8 | Hurt |
| Remove task-level covariance regularization (ablation) | Simple CNAPS · Meta-Dataset overall average | accuracy: 72.2±0.8 -> 72.0±0.8 | Hurt |
| Remove feature extractor adaptation (ablation) | Simple CNAPS · Meta-Dataset overall average | accuracy: 72.2±0.8 -> 59.3±0.9 | Hurt |
| Add a learned projection network (ablation) | Simple CNAPS · Meta-Dataset in-domain average | accuracy: 73.8±0.8 -> 72.4±0.9 | No clear effect |
| Add a learned projection network (ablation) | Simple CNAPS · Meta-Dataset out-of-domain average | accuracy: 69.7±0.8 -> 67.1±0.8 | Hurt |
| Use regularized Mahalanobis distance instead of squared Euclidean distance (ablation) | Simple CNAPS · Meta-Dataset overall average | accuracy: 69.6±0.8 -> 72.2±0.8 | Helped |
| Use regularized Mahalanobis distance instead of cosine similarity (ablation) | Simple CNAPS · Meta-Dataset overall average | accuracy: 68.3±0.8 -> 72.2±0.8 | Helped |
| Use regularized Mahalanobis distance instead of absolute distance (ablation) | Simple CNAPS · Meta-Dataset overall average | accuracy: 68.8±0.8 -> 72.2±0.8 | Helped |
| Use regularized Mahalanobis distance instead of negative dot product (ablation) | Simple CNAPS · Meta-Dataset overall average | accuracy: 61.9±0.9 -> 72.2±0.8 | Helped |
| Use class-specific covariance with task-level covariance regularization (ablation) | Simple CNAPS · Meta-Dataset overall average | accuracy: 72.0±0.8 -> 72.2±0.8 | Helped |
| Use task-adapted feature extraction with Mahalanobis classification (ablation) | Simple CNAPS · Meta-Dataset overall average | accuracy: 59.3±0.9 -> 72.2±0.8 | Helped |
Closest papers by meaning
- A Closer Look at Few-shot Classification Again (ICML 2023 · similarity 0.90)
- Dynamic Few-Shot Visual Learning Without Forgetting (CVPR 2018 · similarity 0.89)
- Joint Distribution Matters: Deep Brownian Distance Covariance for Few-Shot Classification (CVPR 2022 · similarity 0.89)
- A Closer Look at Few-shot Classification (ICLR 2019 · similarity 0.89)
- Few-Class Arena: A Benchmark for Efficient Selection of Vision Models and Dataset Difficulty Measurement (ICLR 2025 · similarity 0.88)
- Discriminative Sample-Guided and Parameter-Efficient Feature Space Adaptation for Cross-Domain Few-Shot Learning (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.