bayesian-optimization

This paper introduces an agentic calibration method that uses a large language model as an optimizer for calibrating grey-box simulation models in cost-effectiveness analysis. The LLM-driven approach achieves competitive performance with substantially fewer model evaluations compared to traditional methods like Nelder-Mead and Bayesian optimization.

This paper introduces ALAS, a flexible Gaussian Process kernel family that learns the stability parameter from data to adapt smoothness, capturing both smooth trends and sharp irregularities, with a separable variant for higher dimensions and theoretical guarantees on information gain.

This paper introduces Maximally Robust Satisficing Bayesian Optimization (MRSBO), a method that efficiently finds solutions meeting a quality threshold while being robust to input perturbations after deployment, outperforming previous approaches.

This paper identifies two structural mechanisms causing multi-task Gaussian processes to misestimate cross-task correlation in Bayesian optimization transfer learning, even for affinely related tasks. The authors propose three conservative remedies to mitigate these issues.

The paper extends the LLaMEA framework to automatically design multi-objective Bayesian optimization algorithms using large language models as mutation and crossover operators within evolutionary strategies, achieving state-of-the-art accuracy with significantly lower computational cost on synthetic and real-world problems.

This paper presents a multi-objective Bayesian optimization approach to automate weight selection in reinforcement learning for energy-aware control, demonstrating superior sample efficiency over grid search on a physical Quanser Aero 2 testbed.

Introduces Bootstrap Flow-Map Tree (BFMT), a computationally efficient sampling framework for history-aware global search and alignment under budget constraints, enabling dynamic transition from exploration to refinement.

This paper presents a method to optimize GPT-5.2 temperature for Naranjo causality assessment in pharmacovigilance, achieving significant agreement improvements via Bayesian hyperparameter optimization with a novel composite metric (EWACS).

This paper presents a range-aware Bayesian optimization framework that directly scores the posterior probability that a candidate satisfies a target property range, enabling discovery of diverse valid designs across multiple specifications.

This paper demonstrates that Optuna's constrained Tree-Structured Parzen Estimator (TPE) is a joint density generalization of the c-TPE algorithm, showing its invariance to constraint duplication while independent c-TPE degrades. The authors outline practical tradeoffs and directions for future study.

This paper identifies the geometric mechanism behind boundary-induced acquisition bias in Gaussian processes on bounded domains, showing how kernel truncation inflates posterior variance and distorts acquisition functions independently of the objective function. The authors introduce a function-free diagnostic to quantify this bias across different acquisition classes.

ReLoRA is a knowledge-reusing adaptation framework that efficiently restores service-ready LoRA adapters for evolving LLM services, reducing time-to-readiness by up to 8.9× and improving accuracy by up to 4.6% through adaptive initialization and scheduled regularization.

The paper proposes a permutation-invariant Bayesian optimization method based on Optimal Transport for optimizing offshore wind farm layouts, which reduces computation time by half and yields better layouts compared to vanilla Bayesian optimization.

The paper introduces Kernel Discovery, an LLM-driven evolutionary framework for high-dimensional Bayesian optimization that searches a broader kernel space and achieves state-of-the-art results on benchmarks.

The LEAP framework integrates a domain-specialized large language model with active learning to efficiently prioritize precursor additives for perovskite solar cells, achieving improved power conversion efficiencies in experimental validation.

This paper provides a unified theoretical framework for pseudo observation batch Bayesian optimization, proving that Gaussian processes produce distinct batch points and that common methods like Constant Liar and Kriging Believer are instances of a single conditioning mechanism. It introduces the Structural Diversity Diagnostic (SDD) for testing surrogate compatibility and validates predictions across multiple benchmark functions and hyperparameter tuning.

This paper formalizes trust calibration for agentic tool use as a preference learning problem, using Gaussian processes and Bayesian optimization to decide when an AI agent's actions should be autonomous or require human approval.

Introduces ReElicit, a Bayesian optimization framework that uses LLMs to elicit and adapt feature spaces for optimizing system prompts under aggregate scalar feedback, achieving strong performance across ten benchmark tasks.

The paper introduces LEAPBench, a 55-task framework for trajectory-level evaluation of LLMs in iterative scientific design, revealing that outcome-based scoring misses efficiency gains and that domain-agnostic prompting can outperform domain-aware prompting in matching published best designs.