Research map / Probabilistic and scientific modeling / Bayesian optimization
Regret bounds for Bayesian optimization
Prove regret guarantees for Bayesian and preferential optimization in kernel bandits.
43 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Bayesian optimization, keyword cluster “acquisition · bayesian · multi objective”
State History Tried Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
15 papers in 2023–24 and 11 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 16.3% to 8.2%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Minimum Regret Search for Single- and Multi-Task Optimization (ICML 2016 · 11 citations)
- 2017 · Counterfactual Data-Fusion for Online Reinforcement Learners (ICML 2017 · 41 citations)
- 2018 · Scalar Posterior Sampling with Applications (NeurIPS 2018 · 4 citations)
- 2019 · Myopic Posterior Sampling for Adaptive Goal Oriented Design of Experiments (ICML 2019 · 13 citations)
- 2020 · Bayesian Optimization for Categorical and Category-Specific Continuous Inputs (AAAI 2020 · 45 citations)
- 2021 · Bayesian Optimistic Optimisation with Exponentially Decaying Regret (ICML 2021 · 0 citations)
- 2022 · Movement Penalized Bayesian Optimization with Application to Wind Energy Systems (NeurIPS 2022 · 4 citations)
- 2023 · Meta Learning of Interface Conditions for Multi-Domain Physics-Informed Neural Networks (ICML 2023 · 5 citations)
- 2024 · Learning Encodings for Constructive Neural Combinatorial Optimization Needs to Regret (AAAI 2024 · 6 citations)
- 2025 · Context in Public Health for Underserved Communities: A Bayesian Approach to Online Restless Bandits (AAAI 2025 · 1 citation)
- 2026 · Divide and Learn: Multi-Objective Combinatorial Optimization at Scale (ICML 2026 · 0 citations)
Earliest papers here
- Minimum Regret Search for Single- and Multi-Task Optimization (ICML 2016 · 11 citations)
- Counterfactual Data-Fusion for Online Reinforcement Learners (ICML 2017 · 41 citations)
- Scalar Posterior Sampling with Applications (NeurIPS 2018 · 4 citations)
What has been tried
35 papers here report 269 measured changes (a component added, removed or swapped, compared with a baseline): helped 65%, no clear effect 11%, hurt 17%, mixed 7%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add tree search to NeuralTS-DE | AAV, TEM, and AAYL49 · maximum fitness | Helped | Tree Search-Based Evolutionary Bandits for Protein Sequence Optimization (AAAI 2024 · 1 citation) | |
| Use exponential rather than standard WRS weights | 16 benchmark datasets · ROP | PAC: 14 vs. 13; FSOL: 13 vs. 16 | Hurt | Stabilizing Linear Passive-Aggressive Online Learning with Weighted Reservoir Sampling (NeurIPS 2024 · 0 citations) |
| Vary the sampling input among ¯xt ▽ ¯x′t, ¯xt △ ¯x′t, ¯xt ∨ ¯x′t, and ¯xt ∧ ¯x′t | Top-5 set: GP sample, Branin-Hoo, Goldstein-Price, Hartmann-6D, NO3, Phosphorus, and Humidity · regret | No clear effect | Active Set Ordering (NeurIPS 2024 · 0 citations) | |
| Train ExPerior-MaxEnt with β = 10 across expert types | Bernoulli bandits, misspecified expert models · Bayesian regret | 51.7 ± 5.1 -> 29.7 ± 3.6 | Helped | Sequential Decision Making with Expert Demonstrations under Unobserved Heterogeneity (NeurIPS 2024 · 0 citations) |
| Vary the growth function to g(t) = max(t₀, tᵃ), with a in (0.25, 0.5, 0.75) | Michalewicz · Best regret | No clear effect | Bayesian Optimisation with Unknown Hyperparameters: Regret Bounds Logarithmically Closer to Optimal (NeurIPS 2024 · 0 citations) |
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 26 papers; from OpenAlex and the papers' affiliations).
- National University of Singapore · 5 papers
- Massachusetts Institute of Technology · 3 papers
- Harvard University · 2 papers
- MediaTek Research · 2 papers
- MedaiTek Research · 2 papers
- Temasek Life Sciences Laboratory / National University of Singapore · 2 papers
- University of Oxford · 2 papers
Use it
Download BibTeX of 35 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Acquisition function design (97 papers)
- Local and high-dimensional Bayesian optimization (90 papers)
- Multi-fidelity hyperparameter optimization (44 papers)
- Multi-objective Bayesian optimization (33 papers)
Elsewhere in the topic
- Offline black-box optimization with surrogates (56 papers)
- High-dimensional Bayesian optimization (34 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).