Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
The Effect of Diversity in Meta-Learning
Ramnath Kumar, Tristan Deleu, Yoshua Bengio
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| Limit task diversity by reusing a task set (ablation) | Meta-learning models · Few-shot classification tasks across Omniglot, miniImageNet, tieredImageNet, and Meta-Dataset | accuracy | No clear effect |
| Use a low-diversity sampler with tasks repeated within batches rather than a single-batch uniform sampler (ablation) | Meta-learning models · Few-shot classification | accuracy | Improved |
| Increase task diversity using OHTM or DPP sampling (ablation) | Meta-learning models · Few-shot classification tasks across the studied datasets | accuracy | No clear effect |
| Increase task diversity using the dynamic DPP sampler (ablation) | Meta-learning models · Few-shot classification | accuracy | Hurt |
| Use OHTM sampling (ablation) | Meta-learning models, especially metric-based methods · Few-shot classification | accuracy | Mixed |
Closest papers by meaning
- Towards Task Sampler Learning for Meta-Learning (IJCV 2024 · similarity 0.90)
- Meta-Learning of Structured Task Distributions in Humans and Machines (ICLR 2021 · similarity 0.89)
- Learning to Learn with Contrastive Meta-Objective (NeurIPS 2025 · similarity 0.86)
- When can in-context learning generalize out of task distribution? (ICML 2025 · similarity 0.85)
- Task Groupings Regularization: Data-Free Meta-Learning with Heterogeneous Pre-trained Models (ICML 2024 · similarity 0.85)
- Theoretical Investigations and Practical Enhancements on Tail Task Risk Minimization in Meta Learning (NeurIPS 2024 · 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.
