On the Correspondence between Gaussian Processes and Geometric Harmonics

Abstract

We discuss the correspondence between Gaussian process regression and Geometric Harmonics, two similar kernel-based methods that are typically used in different contexts. Research communities surrounding the two concepts often pursue different goals. Results from both camps can be successfully combined, providing alternative interpretations of uncertainty in terms of error estimation, or leading towards accelerated Bayesian Optimization due to dimensionality reduction.

arXiv comment26 pages, 9 figures

References in corpus (10)

• A Tutorial on Bayesian Optimization
• Molecular enhanced sampling with autoencoders: On-the-fly collective variable discovery and accelerated free energy landscape exploration
• Understanding and mitigating gradient pathologies in physics-informed neural networks
• Gaussian Process Regression for Geometry Optimization
• Neural-net-induced Gaussian process regression for function approximation and PDE solution
• Probabilistic learning on manifolds constrained by nonlinear partial differential equations for small datasets
• On the Koopman operator of algorithms
• Conditional Karhunen-Loève expansion for uncertainty quantification and active learning in partial differential equation models
• Selecting the independent coordinates of manifolds with large aspect ratios
• Streaming k-PCA: Efficient guarantees for Oja's algorithm, beyond rank-one updates