Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
MATE: Plugging in Model Awareness to Task Embedding for 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 |
|---|---|---|---|
| add MATE to ProtoNets (ResNet-12) | ProtoNets · CIFAR-FS 5-way 5-shot | accuracy: 84.07 -> 84.71 | Helped |
| add MATE to ProtoNets (ResNet-12) | ProtoNets · CIFAR-FS 5-way 1-shot | accuracy: 71.35 -> 71.49 | No clear effect |
| add MATE to R2D2 (ResNet-12) | R2D2 · CIFAR-FS 5-way 5-shot | accuracy: 84.60 -> 85.04 | Helped |
| add MATE to R2D2 (ResNet-12) | R2D2 · CIFAR-FS 5-way 1-shot | accuracy: 72.51 -> 72.59 | No clear effect |
| add MATE to MetaOptNet | MetaOptNet · CIFAR-FS 5-way 5-shot | accuracy: 84.2 -> 85.2 | Helped |
| add MATE to MetaOptNet | MetaOptNet · CIFAR-FS 5-way 1-shot | accuracy: 72.0 -> 72.3 | No clear effect |
| add MATE to MetaOptNet (vs replicated baseline) | MetaOptNet · miniImageNet 5-way 5-shot | accuracy: 77.88 -> 78.64 | Helped |
| add MATE to MetaOptNet (vs replicated baseline) | MetaOptNet · miniImageNet 5-way 1-shot | accuracy: 61.64 -> 62.08 | Helped |
| add MATE to MetaOptNet (vs originally reported numbers) | MetaOptNet · miniImageNet 5-way 1-shot | accuracy: 62.64 -> 62.08 | Hurt |
| replace concatenation with FiLM fusion (both with KME) (ablation) | MATE + MetaOptNet · CIFAR-FS 5-way 1-shot | accuracy: 71.41 -> 71.55 | No clear effect |
| replace plain KME with SVM-based feature attention (with FiLM) (ablation) | MATE + MetaOptNet · CIFAR-FS 5-way 1-shot | accuracy: 71.55 -> 72.15 | Helped |
| replace plain KME with SVM-based feature attention (with FiLM) (ablation) | MATE + MetaOptNet · CIFAR-FS 5-way 5-shot | accuracy: 84.37 -> 84.72 | Helped |
Closest papers by meaning
- Efficient and Effective Multi-task Grouping via Meta Learning on Task Combinations (NeurIPS 2022 · similarity 0.87)
- Bridging Multi-Task Learning and Meta-Learning: Towards Efficient Training and Effective Adaptation (ICML 2021 · similarity 0.87)
- Hierarchical Gaussian Mixture based Task Generative Model for Robust Meta-Learning (NeurIPS 2023 · similarity 0.87)
- Task Groupings Regularization: Data-Free Meta-Learning with Heterogeneous Pre-trained Models (ICML 2024 · similarity 0.86)
- Task-Robust Model-Agnostic Meta-Learning (NeurIPS 2020 · similarity 0.86)
- Unsupervised Meta-Learning via In-Context Learning (ICLR 2025 · similarity 0.85)
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.