Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Learning To Affiliate: Mutual Centralized Learning for Few-Shot 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 unidirectional matching with bidirectional random walks (ablation) | MCL with VanillaFCN · miniImageNet 1-shot, Conv-4 | accuracy (%): 54.88 -> 55.38 | Helped |
| replace unidirectional matching with bidirectional random walks (ablation) | MCL with VanillaFCN · tieredImageNet 5-shot, Conv-4 | accuracy (%): 68.87 -> 74.25 | Helped |
| apply Katz attenuation with alpha=0.5 instead of eigenvector centrality approximation (ablation) | MCL with VanillaFCN · miniImageNet 5-shot, Conv-4 | accuracy (%): 70.02 -> 71.74 | Helped |
| apply centrality pooling to query features only (ablation) | ProtoNet · miniImageNet 1-shot, Conv-4 | accuracy (%): 52.32 -> 53.74 | Helped |
| apply centrality pooling to support features only (ablation) | ProtoNet · tieredImageNet 1-shot, Conv-4 | accuracy (%): 53.19 -> 56.60 | Helped |
| apply centrality pooling to both query and support features (ablation) | ProtoNet · miniImageNet 1-shot, Conv-4 | accuracy (%): 52.32 -> 54.31 | Helped |
| vary Katz attenuation alpha from the unidirectional limit to 0.5 (ablation) | MCL-Katz with VanillaFCN, ResNet-12 · miniImageNet 5-way 1-shot | accuracy (%): 66.60 -> 67.51 | Helped |
| vary Katz attenuation alpha from the unidirectional limit to 0.7 (ablation) | MCL-Katz with VanillaFCN, ResNet-12 · miniImageNet 5-way 5-shot | accuracy (%): 81.76 -> 84.05 | Helped |
| increase gamma and beta beyond the selected setting (ablation) | MCL · miniImageNet 5-way 1-shot, VanillaFCN | accuracy (%): 67.36 -> 65.08 | Hurt |
| use centrality-weighted pooling in place of global average pooling | ProtoNet · miniImageNet 1-shot, Conv-4 | accuracy (%): 52.32 -> 54.31 | Helped |
| use centrality-weighted pooling in place of global average pooling | RelationNet · tieredImageNet 1-shot, Conv-4 | accuracy (%): 54.33 -> 57.73 | Helped |
Closest papers by meaning
- DMN4: Few-Shot Learning via Discriminative Mutual Nearest Neighbor Neural Network (AAAI 2022 · similarity 0.88)
- Dense Classification and Implanting for Few-Shot Learning (CVPR 2019 · similarity 0.88)
- Pushing the Limits of Simple Pipelines for Few-Shot Learning: External Data and Fine-Tuning Make a Difference (CVPR 2022 · similarity 0.88)
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.88)
- Discriminative Sample-Guided and Parameter-Efficient Feature Space Adaptation for Cross-Domain Few-Shot Learning (CVPR 2024 · similarity 0.87)
- MOKD: Cross-domain Finetuning for Few-shot Classification via Maximizing Optimized Kernel Dependence (ICML 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.