Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Bridging Multi-Task Learning and Meta-Learning: Towards Efficient Training and Effective Adaptation
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 MTL instead of MetaOptNet | Few-shot image classification · mini-ImageNet 5-way | accuracy: 62.64 -> 59.84; 78.63 -> 77.72 | Mixed |
| Use MTL instead of MetaOptNet | Few-shot image classification · tiered-ImageNet 5-way | accuracy: 65.99 -> 67.11; 81.56 -> 83.69 | Helped |
| Use MTL instead of MetaOptNet | Few-shot image classification · CIFAR-FS 5-way | accuracy: 72.6 -> 69.5; 84.3 -> 84.1 | Mixed |
| Use MTL instead of MetaOptNet | Few-shot image classification · FC100 5-way | accuracy: 41.1 -> 42.4; 55.5 -> 57.7 | Helped |
| Use first-order MTL training instead of MetaOptNet training | mini-ImageNet 5-way 5-shot · mini-ImageNet | GPU Hours: 85.6 hrs -> 3.7 hrs | Helped |
| Use MTL rather than MetaOptNet, with reduced training time | tiered-ImageNet 5-way 5-shot · tiered-ImageNet | accuracy: 81.55% vs. 81.56% | No clear effect |
| Decrease the inner-loop product λτ | MTL and ANIL predictors · Synthetic few-shot regression | prediction difference | Helped |
| Increase network depth L | MTL and ANIL predictors · Synthetic few-shot regression | prediction difference | Helped |
| Add feature normalization (ablation) | MTL few-shot classification · mini-ImageNet, tiered-ImageNet, and CIFAR-FS | accuracy | Helped |
| Add feature normalization (ablation) | MTL few-shot classification · FC100 | accuracy | No clear effect |
Closest papers by meaning
- Efficient and Effective Multi-task Grouping via Meta Learning on Task Combinations (NeurIPS 2022 · similarity 0.91)
- Meta-Learning without Memorization (ICLR 2020 · similarity 0.91)
- Multi-Task Learning with User Preferences: Gradient Descent with Controlled Ascent in Pareto Optimization (ICML 2020 · similarity 0.90)
- Large-Scale Meta-Learning with Continual Trajectory Shifting (ICML 2021 · similarity 0.89)
- Delve into the Applicability of Advanced Optimizers for Multi-Task Learning (NeurIPS 2026 · similarity 0.89)
- Task Groupings Regularization: Data-Free Meta-Learning with Heterogeneous Pre-trained Models (ICML 2024 · similarity 0.89)
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.