Research map / Architectures and efficiency / Neural architecture search
HyperJump: Accelerating HyperBand via Risk Modelling
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 |
|---|---|---|---|
| Skip low-risk HyperBand stages using risk modelling | HyperJump · NATS-Bench ImageNet-16-120 and Cifar100, sequential | speedup to recommend near-optimal configurations: around 20× to 32× | Helped |
| Skip low-risk HyperBand stages using risk modelling | HyperJump · LIBSVM Covertype, sequential | time to reach loss around 0.08: approx. 7× | Helped |
| Use HyperJump rather than HyperBand | LIBSVM Covertype, sequential · LIBSVM Covertype | final loss: 0.072±3.6E-3 -> 0.066±8.9E-4 | Helped |
| Use HyperJump rather than BOHB | LIBSVM Covertype, sequential · LIBSVM Covertype | final loss: 0.72±3.5E-3 -> 0.066±8.9E-4 | Helped |
| Skip low-risk HyperBand stages using risk modelling | HyperJump · NATS-Bench ImageNet-16-120 and Cifar100, 32 workers | speedup to recommend near-optimal configurations: up to approximately 6× | Helped |
| Add the jumping logic (ablation) | HyperJump · ImageNet, sequential | optimization performance | Helped |
| Add configuration evaluation-order prioritization (ablation) | HyperJump · ImageNet, sequential | optimization performance | Helped |
| Add pause-resume training and opportunistic evaluation (ablation) | HyperJump · ImageNet, sequential | optimization performance | No clear effect |
| Add bracket warm-starting (ablation) | HyperJump · ImageNet, sequential | optimization performance | No clear effect |
Closest papers by meaning
- Hyperband: Bandit-Based Configuration Evaluation for Hyperparameter Optimization (ICLR 2017 · similarity 0.90)
- BOHB: Robust and Efficient Hyperparameter Optimization at Scale (ICML 2018 · similarity 0.87)
- MFES-HB: Efficient Hyperband with Multi-Fidelity Quality Measurements (AAAI 2021 · similarity 0.87)
- Sample complexity of data-driven tuning of model hyperparameters in neural networks with structured parameter-dependent dual function (NeurIPS 2025 · similarity 0.85)
- Efficient Hyperparameter Optimization with Adaptive Fidelity Identification (CVPR 2024 · similarity 0.85)
- Cost-Sensitive Freeze-thaw Bayesian Optimization for Efficient Hyperparameter Tuning (NeurIPS 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.