arXiv：cs.LG· Quoc-Anh Hoang Nguyen, Xinyu Zhang, Daolang Huang, Julien Martinelli, Samuel Kaski·· 3 小时前
FMTR：基于 Flow Matching 的上下文批量黑盒优化
In-Context Batch Black-box Optimization via Flow Matching
AI 导读
研究者提出 Flow Matching for Thompson-Sampling Refinement（FMTR），一种摊销式批量黑盒优化方法，通过条件流匹配离线学习历史条件下的连续候选分布，可采样可变大小的批量。测试阶段模型参数固定，无需重训任务专属代理模型或做数值采集优化。FMTR 在合成与真实基准上优于批量贝叶斯优化及上下文基线，并大幅降低批量选择计算成本。
正文
Abstract:Batch black-box optimization is important for expensive systems where many candidates can be evaluated in parallel, yet only a few rounds of evaluation are affordable. In this regime, effective batch construction must identify multiple promising regions while avoiding redundant evaluations, and conventional Bayesian optimization can require increasingly difficult batch-selection procedures as the batch size grows. We introduce Flow Matching for Thompson-Sampling Refinement (FMTR), an amortized approach that learns offline a continuous, history-conditioned proposal distribution from which variable-size batches can be sampled. FMTR is trained offline by conditional flow matching on Thompson-sampling argmaxes refined toward high-value regions via a horizon-dependent Langevin pushforward map, adapting the exploration-exploitation tradeoff over the optimization horizon. At test time, the model parameters remain fixed and batch generation requires neither task-specific surrogate refitting nor numerical acquisition optimization. FMTR achieves strong performance against batch BO and in-context baselines on synthetic and real-world benchmarks, while substantially reducing the computational cost of batch selection.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.20138 [cs.LG] |
| (or arXiv:2610.20138v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.20138 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Quoc-Anh Nguyen Hoang [view email]
[v1]
Wed, 14 Oct 2026 12:55:43 UTC (16,108 KB)
来源：arXiv：cs.LG · arxiv.org