arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2402.02746cs.LGstat.ML
收藏

标准高斯过程足以胜任高维贝叶斯优化

Standard Gaussian Process is All You Need for High-Dimensional Bayesian Optimization

Zhitong Xu, Haitao Wang, Jeff M Phillips, Shandian Zhe

AI总结:

本文通过实证与理论分析证明标准高斯过程贝叶斯优化在高维问题中表现不佳主要源于SE核的长度尺度初始化不当,并提出一种简单稳健的初始化策略,使标准BO达到接近最先进方法的性能。

摘要:

一个长期存在的观点认为,使用标准高斯过程(GP)的贝叶斯优化(BO)——称为标准BO——在高维优化问题中表现不佳。尽管这一观点看似合理,但缺乏有力的经验证据和理论依据。为填补这一空白,我们进行了系统性研究。首先,通过在十二个基准上的全面评估,我们发现虽然流行的平方指数(SE)核常常导致性能不佳,但使用Matérn核可使标准BO持续取得一流结果,经常超越专门为高维优化设计的方法。其次,我们的理论分析揭示,SE核的失败主要源于长度尺度参数的不当初始化,这种初始化在实践中常用,但可能导致训练中的梯度消失。我们提供了一个概率界来刻画这一问题,表明Matérn核受此影响较小,能够稳健地处理高得多的维度。第三,我们提出了一种简单的稳健初始化策略,显著提升了SE核的性能,使其接近最先进方法,且无需额外先验或正则化。我们证明了另一个概率界,展示该方法如何有效缓解梯度消失问题。我们的发现主张重新评估标准BO在高维场景下的潜力。

英文摘要:

A long-standing belief holds that Bayesian Optimization (BO) with standard Gaussian processes (GP) -- referred to as standard BO -- underperforms in high-dimensional optimization problems. While this belief seems plausible, it lacks both robust empirical evidence and theoretical justification. To address this gap, we present a systematic investigation. First, through a comprehensive evaluation across twelve benchmarks, we found that while the popular Square Exponential (SE) kernel often leads to poor performance, using Matérn kernels enables standard BO to consistently achieve top-tier results, frequently surpassing methods specifically designed for high-dimensional optimization. Second, our theoretical analysis reveals that the SE kernel's failure primarily stems from improper initialization of the length-scale parameters, which are commonly used in practice but can cause gradient vanishing in training. We provide a probabilistic bound to characterize this issue, showing that Matérn kernels are less susceptible and can robustly handle much higher dimensions. Third, we propose a simple robust initialization strategy that dramatically improves the performance of the SE kernel, bringing it close to state-of-the-art methods, without requiring additional priors or regularization. We prove another probabilistic bound that demonstrates how the gradient vanishing issue can be effectively mitigated with our method. Our findings advocate for a re-evaluation of standard BO's potential in high-dimensional settings.

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。

↑