Early-Signal Detector

method · first seen 5 Oct 2026

Bayesian optimization

Also: BO, Bayesian optimisation, Bayesian data augmentation

Sequential model-based optimization using a surrogate model to pick informative evaluations.

Bayesian optimization fits a probabilistic surrogate, usually a Gaussian process, to expensive black-box evaluations and uses an acquisition function to choose the next query. The sources are mostly arXiv papers applying or refining this long-standing method rather than introducing it. They include isotropic GP priors for high-dimensional BO, trust-region molecular design (TRACE), tabular foundation models such as TabPFN as surrogates in materials discovery, LLM-guided configuration of BO (STaSBO), and BO set against agentic optimization in self-driving lab testbeds. There is also one protein-engineering repo. The spread across chemistry, materials, NAS and protein design reflects routine use of the method.

The method itself has been mainstream for over a decade, so the rise is noise. The more specific threads are worth watching separately: foundation-model surrogates, LLM meta-controllers for BO, and BO inside self-driving laboratories.

Analyst note (established), 17 Nov 2026.

Mentions per week

01419 May11 Aug10 NovPapersCodeDiscussion

PapersCodeDiscussion

Rank 135 on 14 Nov 2026 · 36 mentions in 28 days · 34 independent groups · papers → code · first seen 6 weeks ago · acceleration +0.27

Earliest sources

Latest sources

Mentioned together with

Gaussian processself-driving laboratorylow-rank adaptationtrust-region methodvariational autoencoderGaussian process banditsactive learningLLM-guided Bayesian optimizationhyperparameter optimizationdirected evolution