Research map / Optimization / Optimizers
Bayesian Optimization Meets Bayesian Optimal Stopping
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 |
|---|---|---|---|
| Increasing K1,t sequence (penalty for early stopping grows over BO iterations) vs fixed K1 (ablation) | BO-BOS on logistic regression (MNIST) · MNIST logistic regression hyperparameter tuning | best-found validation error | Helped |
| Adding BO-BOS early stopping (BOS with GP-UCB) vs plain GP-UCB | CNN hyperparameter tuning on CIFAR-10 and SVHN · CIFAR-10, SVHN | best-found validation error vs run-time | Helped |
| Hyperband instead of BO-BOS | CNN hyperparameter tuning · CIFAR-10, SVHN | best-found validation error | Hurt |
| Using BOCA multi-fidelity instead of BO-BOS | CNN hyperparameter tuning · CIFAR-10, SVHN | best-found validation error | Hurt |
| LC prediction (learning curve extrapolation) added to GP-UCB | CNN hyperparameter tuning · CIFAR-10, SVHN | best-found validation error | No clear effect |
| Discounting rewards (gamma=0.9) and N=50 epochs for BO-BOS vs discount 1.0 (undiscounted) (ablation) | BO-BOS policy search on Swimmer-v2 · Swimmer-v2 RL | best-found un-discounted return | Helped |
| Hyperband instead of BO-BOS for policy search with 16 parameters | Hyperband on Swimmer-v2 policy search · Swimmer-v2 RL | average return | Hurt |
| Adding BO-BOS early stopping for joint hyperparameter tuning and feature selection (forward selection as epochs) | XGBoost on email spam dataset · joint tuning and feature selection | best-found validation error vs run-time | Helped |
| Hyperband instead of BO-BOS for joint tuning and feature selection | XGBoost feature selection · email spam dataset | best-found validation error vs run-time | Hurt |
Closest papers by meaning
- Bayesian Optimisation with Unknown Hyperparameters: Regret Bounds Logarithmically Closer to Optimal (NeurIPS 2024 · similarity 0.91)
- Posterior Sampling-Based Bayesian Optimization with Tighter Bayesian Regret Bounds (ICML 2024 · similarity 0.89)
- Efficient Hyperparameter Optimization with Adaptive Fidelity Identification (CVPR 2024 · similarity 0.89)
- Understanding High-Dimensional Bayesian Optimization (ICML 2025 · similarity 0.88)
- Minimizing UCB: a Better Local Search Strategy in Local Bayesian Optimization (NeurIPS 2024 · similarity 0.88)
- Batched Energy-Entropy acquisition for Bayesian Optimization (NeurIPS 2024 · similarity 0.88)
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.