arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2407.03608math.NAcs.NAstat.CO
收藏

高斯过程回归中的对数线性缩放用于常见非平稳核

Gaussian process regression with log-linear scaling for common non-stationary kernels

P. Michael Kielstra, Michael Lindsey

AI总结:

本文提出一种适用于低维非平稳核的高斯过程回归快速算法,利用非均匀快速傅里叶变换实现近最优的$O(N\log N)$复杂度,并在多维空间展示优于秩结构方法的可扩展性。

摘要:

我们提出了一种用于低维高斯过程回归的快速算法,适用于一类广泛使用的非平稳核。这些核的非平稳性由任意空间变化的垂直和水平尺度引起。特别地,任何平稳核都可以作为特例被容纳,我们尤其关注标准Matérn核的推广。我们的核矩阵-向量乘法子程序以几乎最优的复杂度$O(N\log N)$进行缩放,其中$N$是回归点的数量。与最近发展的仅适用于平稳核的等距傅里叶高斯过程(EFGP)方法类似,我们的方法利用了非均匀快速傅里叶变换(NUFFTs)。我们提供了控制方法近似误差的完整分析,并通过数值实验验证了方法的实际性能。特别是,我们展示了在空间维度$d>1$下,与最先进的秩结构方法相比,我们的方法具有更好的可扩展性。

英文摘要:

We introduce a fast algorithm for Gaussian process regression in low dimensions, applicable to a widely-used family of non-stationary kernels. The non-stationarity of these kernels is induced by arbitrary spatially-varying vertical and horizontal scales. In particular, any stationary kernel can be accommodated as a special case, and we focus especially on the generalization of the standard Matérn kernel. Our subroutine for kernel matrix-vector multiplications scales almost optimally as $O(N\log N)$, where $N$ is the number of regression points. Like the recently developed equispaced Fourier Gaussian process (EFGP) methodology, which is applicable only to stationary kernels, our approach exploits non-uniform fast Fourier transforms (NUFFTs). We offer a complete analysis controlling the approximation error of our method, and we validate the method's practical performance with numerical experiments. In particular we demonstrate improved scalability compared to to state-of-the-art rank-structured approaches in spatial dimension $d>1$.

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。

↑