Research map / Learning theory and algorithms / Prediction and decision losses
When Do Curricula Work?
Xiaoxia Wu, Ethan Dyer, Behnam Neyshabur
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 |
|---|---|---|---|
| Use curriculum ordering instead of standard i.i.d. training | ResNet-50 · CIFAR10 and CIFAR100, standard training time | test accuracy | No clear effect |
| Use anti-curriculum ordering instead of standard i.i.d. training | ResNet-50 · CIFAR10 and CIFAR100, standard training time | test accuracy | No clear effect |
| Use random ordering with a dynamically expanding training set instead of standard i.i.d. training | ResNet-50 · CIFAR10 and CIFAR100, standard training time | test accuracy | No clear effect |
| Use curriculum ordering instead of standard i.i.d. training | ResNet-50 · FOOD101 and FOOD101N, standard training time | test accuracy | No clear effect |
| Use curriculum ordering instead of standard i.i.d. training | ResNet-50 · CIFAR10 and CIFAR100, 1760 or 352 training steps | test accuracy | Improved |
| Use anti-curriculum ordering instead of standard i.i.d. training | ResNet-50 · CIFAR10 and CIFAR100, 1760 or 352 training steps | test accuracy | No clear effect |
| Use random ordering with a dynamically expanding training set instead of standard i.i.d. training | ResNet-50 · CIFAR10 and CIFAR100, 352 training steps | test accuracy | Improved |
| Use curriculum ordering instead of standard i.i.d. training | ResNet-50 · CIFAR100 with 20%, 40%, 60%, or 80% label noise | test accuracy | Improved |
| Use curriculum ordering instead of anti-curriculum or random-curriculum ordering | ResNet-50 · FOOD101, limited training time | test accuracy | Improved |
| Use curriculum ordering instead of anti-curriculum or random-curriculum ordering | ResNet-50 · FOOD101N | test accuracy | Improved |
Closest papers by meaning
- An Analytical Theory of Curriculum Learning in Teacher-Student Networks (NeurIPS 2022 · similarity 0.89)
- Learning to Learn: How to Continuously Teach Humans and Machines (ICCV 2023 · similarity 0.88)
- Learning to Solve Complex Problems via Dataset Decomposition (NeurIPS 2025 · similarity 0.85)
- CurBench: Curriculum Learning Benchmark (ICML 2024 · similarity 0.83)
- Montessori-Instruct: Generate Influential Training Data Tailored for Student Learning (ICLR 2025 · similarity 0.83)
- K12-KGraph: A Curriculum-Aligned Knowledge Graph for Benchmarking and Training Educational LLMs (NeurIPS 2026 · similarity 0.83)
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.
