arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2403.04764cs.LGmath.OCstat.ML
收藏

TS-RSR:一种可证明高效的批量贝叶斯优化方法

TS-RSR: A provably efficient approach for batch Bayesian Optimization

Harvard University(哈佛大学)

Zhaolin Ren, Na Li

AI总结:

提出 TS-RSR 批量贝叶斯优化方法,通过最小化遗憾与不确定性之比的 Thompson Sampling 近似协调批量采样,并给出遗憾收敛保证和优于基准算法的实验结果。

摘要:

本文提出了一种用于批量贝叶斯优化(BO)的新方法,称为 Thompson Sampling-Regret to Sigma Ratio directed sampling(TS-RSR)。该方法通过最小化遗憾与不确定性之比的 Thompson Sampling 近似,来采样一批新的动作。我们的采样目标能够协调每一批中选择的动作,从而最小化各点之间的冗余,同时关注具有较高预测均值或较高不确定性的点。理论上,我们为算法的遗憾提供了严格的收敛保证;数值上,我们证明该方法在一系列具有挑战性的合成测试函数和真实测试函数上达到了最先进性能,并优于若干具有竞争力的基准批量 BO 算法。

英文摘要:

This paper presents a new approach for batch Bayesian Optimization (BO) called Thompson Sampling-Regret to Sigma Ratio directed sampling (TS-RSR), where we sample a new batch of actions by minimizing a Thompson Sampling approximation of a regret to uncertainty ratio. Our sampling objective is able to coordinate the actions chosen in each batch in a way that minimizes redundancy between points whilst focusing on points with high predictive means or high uncertainty. Theoretically, we provide rigorous convergence guarantees on our algorithm's regret, and numerically, we demonstrate that our method attains state-of-the-art performance on a range of challenging synthetic and realistic test functions, where it outperforms several competitive benchmark batch BO algorithms.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑