Large-Scale Gaussian Processes via Alternating Projection

arXiv:2310.17137v2 Announce Type: replace
Abstract: Training and inference in Gaussian processes (GPs) require solving linear systems with $n\times n$ kernel matrices. To address the prohibitive $\mathcal{O}(n^3)$ time complexity, recent work has employed fast iterative methods, like conjugate gradients (CG). However, as datasets increase in magnitude, the kernel matrices become increasingly ill-conditioned and still require $\mathcal{O}(n^2)$ space without partitioning. Thus, while CG increases the size of datasets GPs can be trained on, modern datasets reach scales beyond its applicability. In this …

abstract arxiv become complexity cs.lg datasets gaussian processes gps however inference iterative kernel linear processes projection scale stat.ml systems training type via work