mlsanityData to 14 Oct 2026

Improving LLM-based Global Optimization with Search Space Partitioning

arXiv 2505.21372 ↗First posted 27 May 2025cs.LG cs.AIcited by 6

Not on the map. Why?

Abstract (arXiv)

Large Language Models (LLMs) have recently emerged as effective surrogate models and candidate generators within global optimization frameworks for expensive blackbox functions. Despite promising results, LLM-based methods often struggle in high-dimensional search spaces or when lacking domain-specific priors, leading to sparse or uninformative suggestions. To overcome these limitations, we propose HOLLM, a novel global optimization algorithm that enhances LLM-driven sampling by partitioning the search space into promising subregions. Each subregion acts as a ``meta-arm'' selected via a bandit-inspired scoring mechanism that effectively balances exploration and exploitation. Within each selected subregion, an LLM then proposes high-quality candidate points, without any explicit domain knowledge. Empirical evaluation on standard optimization benchmarks shows that HOLLM consistently matches or surpasses leading global optimization methods, while substantially outperforming global LLM-based sampling strategies.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Builds on 33 of its 67 references that are in the map. First posted on arXiv 27 May 2025.

Builds on

33 of 67 references.

References · 33

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2016SGDR: Stochastic Gradient Descent with Warm Restartscited by 5,045
  3. 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,224
  4. 2018Regularized Evolution for Image Classifier Architecture Searchcited by 909
  5. 2023Large Language Models as Optimizerscited by 864
  6. 2018A Tutorial on Bayesian Optimizationcited by 744
  7. 2019Introduction to Multi-Armed Banditscited by 481
  8. 2020NAS-Bench-201: Extending the Scope of Reproducible Neural Architecture Searchcited by 416
  9. 2016The CMA Evolution Strategy: A Tutorialcited by 377
  10. 2017A Downsampled Variant of ImageNet as an Alternative to the CIFAR datasetscited by 377
  11. 2019Scalable Global Optimization via Local Bayesian Optimizationcited by 374
  12. 2021SMAC3: A Versatile Bayesian Optimization Package for Hyperparameter Optimizationcited by 230
21 more
  1. 2021Bayesian Optimization is Superior to Random Search for Machine Learning Hyperparameter Tuning: Analysis of the Black-Box Optimization Challenge 2020cited by 148
  2. 2023Neural Architecture Search: Insights from 1000 Paperscited by 139
  3. 2024Large Language Models to Enhance Bayesian Optimizationcited by 135
  4. 2020Hyper-Parameter Optimization: A Review of Algorithms and Applicationscited by 114
  5. 2020Learning Search Space Partition for Black-box Optimization using Monte Carlo Tree Searchcited by 89
  6. 2021Multi-Objective Bayesian Optimization over High-Dimensional Search Spacescited by 87
  7. 2020HEBO Pushing The Limits of Sample-Efficient Hyperparameter Optimisationcited by 83
  8. 2017Parallel and Distributed Thompson Sampling for Large-scale Accelerated Exploration of Chemical Spacecited by 76
  9. 2024Can large language models explore in-context?cited by 76
  10. 2020An Easy-to-use Real-world Multi-objective Optimization Problem Suitecited by 71
  11. 2023Using Large Language Models for Hyperparameter Optimizationcited by 71
  12. 2019Tabular Benchmarks for Joint Architecture and Hyperparameter Optimizationcited by 58
  13. 2024A Sober Look at LLMs for Material Discovery: Are They Actually Good for Bayesian Optimization Over Molecules?cited by 48
  14. 2015Simple regret for infinitely many armed banditscited by 39
  15. 2024Position: Leverage Foundational Models for Black-Box Optimizationcited by 21
  16. 2017Efficient-UCBV: An Almost Optimal Algorithm using Variance Estimatescited by 20
  17. 2021Triangulation candidates for Bayesian optimizationcited by 11
  18. 2023Optimizing Hyperparameters with Conformal Quantile Regressioncited by 10
  19. 2013Adaptive-treed banditscited by 6
  20. 2021Learning Space Partitions for Path Planningcited by 6
  21. 2024Voronoi Candidates for Bayesian Optimizationcited by 5

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 6.

6 papers.

  1. 2026Can LLMs Beat Classical Hyperparameter Optimization Algorithms? A Study on autoresearchcited by 15
  2. 2026Agentic Bayesian Optimization through Surrogate-Augmented Autoresearchcited by 6
  3. 2026Multi-Objective Hierarchical Optimization with Large Language Modelscited by 1
  4. 2026stratum: A System Infrastructure for Massive Agent-Centric ML Workloadscited by 1
  5. 2026An Open-Source Training Dataset for Foundation Models for Black-box Optimizationcited by 0
  6. 2026Agentic High-Dimensional Bayesian Optimization with Hypothesis- and Evidence-Guided Searchcited by 0