Research map / Architectures and efficiency / Neural architecture search
Learning to Rank Learning Curves
Martin Wistuba, Tejaswini Pedapati
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 LCRankNet for early termination instead of training every configuration to completion | Random NAS in the NASNet search space · CIFAR-100 | Time (GPU hours): 1021 -> 11 | Improved |
| Use LCRankNet for early termination instead of training every configuration to completion | Random NAS in the NASNet search space · CIFAR-100 | Regret: 0.00 -> 0.00 | No clear effect |
| Use LCRankNet for early termination instead of training every configuration to completion | Random NAS in the NASNet search space · CIFAR-10 | Time (GPU hours): 1023 -> 20 | Improved |
| Use LCRankNet for early termination instead of training every configuration to completion | Random NAS in the NASNet search space · Fashion-MNIST | Regret: 0.00 -> 0.10 | Hurt |
| Use LCRankNet for early termination instead of training every configuration to completion | Random NAS in the NASNet search space · Quickdraw | Time (GPU hours): 1045 -> 28 | Improved |
| Add metadata from other datasets (ablation) | LCRankNet · Learning-curve ranking across image datasets | Spearman rank correlation | Improved |
| Add architecture description to the learning-curve representation (ablation) | LCRankNet · Learning-curve ranking | Spearman rank correlation | Improved |
| Add learning-curve information to the architecture-only model (ablation) | LCRankNet · Learning-curve ranking | Spearman rank correlation | Improved |
| Use architecture information alone rather than the full representation (ablation) | LCRankNet · Short learning curves | Spearman rank correlation | Mixed |
| Use learning-curve information alone rather than the full representation (ablation) | LCRankNet · Learning-curve ranking | Spearman rank correlation | Mixed |
| Replace pairwise ranking loss with pointwise L2 loss (ablation) | LCRankNet · Short learning curves | Spearman rank correlation | Improved |
| Replace pairwise ranking loss with pointwise L2 loss (ablation) | LCRankNet · Longer learning curves | Spearman rank correlation | Hurt |
Closest papers by meaning
- PiRank: Scalable Learning To Rank via Differentiable Sorting (NeurIPS 2021 · similarity 0.86)
- Architecture-Aware Learning Curve Extrapolation via Graph Ordinary Differential Equation (AAAI 2025 · similarity 0.85)
- Boosting Order-Preserving and Transferability for Neural Architecture Search: a Joint Architecture Refined Search and Fine-tuning Approach (CVPR 2024 · similarity 0.84)
- Maestro: Uncovering Low-Rank Structures via Trainable Decomposition (ICML 2024 · similarity 0.84)
- MetaOptimize: A Framework for Optimizing Step Sizes and Other Meta-parameters (ICML 2025 · similarity 0.83)
- GradMetaNet: An Equivariant Architecture for Learning on Gradients (NeurIPS 2025 · 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.
