Research map / Probabilistic and scientific modeling / Bayesian optimization
Max-value Entropy Search for Efficient Bayesian Optimization
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 |
|---|---|---|---|
| Use MES-G instead of PES | Bayesian optimization · Eggholder | inference regret: 37.94 ± 26.05 -> 46.56 ± 27.05 | Hurt |
| Use MES-G instead of EST | Bayesian optimization · Shekel | inference regret: 5.57 ± 2.56 -> 5.45 ± 2.07 | Helped |
| Use MES-G instead of EI | Bayesian optimization · Michalewicz | inference regret: 4.80 ± 0.60 -> 4.49 ± 0.51 | Helped |
| Use MES-G instead of PES | Bayesian optimization · Boston housing | inference regret: 1.52 ± 0.32 -> 1.51 ± 0.61 | Helped |
| Use MES-G instead of UCB | Bayesian optimization · Breast cancer | inference regret (%): 3.83 ± 0.01 -> 3.83 ± 0.01 | No clear effect |
| Use MES-R instead of PI | Bayesian optimization · Robot pushing, 4-d action | inference regret: 0.16 ± 0.20 -> 0.16 ± 0.10 | No clear effect |
| Use MES-G instead of ES | Bayesian optimization · Robot pushing, 3-d action | inference regret: 0.62 ± 0.59 -> 0.61 ± 1.26 | Helped |
| Use Gumbel sampling rather than posterior-function random-feature sampling (ablation) | MES · Synthetic functions, 3-dimensional GP | simple and inference regret | Helped |
| Increase the number of sampled maxima from 1 to 10 or 100 (ablation) | MES-G and MES-R · Synthetic functions, 3-dimensional GP | simple and inference regret | No clear effect |
| Increase the number of sampled argmaxes from 1 or 10 to 100 (ablation) | PES · Synthetic functions, 3-dimensional GP | simple and inference regret | Helped |
| Use MES-G instead of MES-R (ablation) | Add-MES · Synthetic additive functions, d=10, 20, 30 | simple regret | Helped |
| Use MES-R instead of MES-G (ablation) | Add-MES · Synthetic additive functions, d=50, 100 | simple regret | Helped |
Closest papers by meaning
- Joint Entropy Search For Maximally-Informed Bayesian Optimization (NeurIPS 2022 · similarity 0.94)
- Sequential and Parallel Constrained Max-value Entropy Search via Information Lower Bound (ICML 2022 · similarity 0.92)
- A Unified Framework for Entropy Search and Expected Improvement in Bayesian Optimization (ICML 2025 · similarity 0.91)
- Principled Preferential Bayesian Optimization (ICML 2024 · similarity 0.86)
- Batched Energy-Entropy acquisition for Bayesian Optimization (NeurIPS 2024 · similarity 0.86)
- Cost-aware Bayesian Optimization via the Pandora's Box Gittins Index (NeurIPS 2024 · similarity 0.86)
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.