Fast methods for training Gaussian processes on large data sets

Abstract

Gaussian process regression (GPR) is a non-parametric Bayesian technique for interpolating or fitting data. The main barrier to further uptake of this powerful tool rests in the computational costs associated with the matrices which arise when dealing with large data sets. Here, we derive some simple results which we have found useful for speeding up the learning stage in the GPR algorithm, and especially for performing Bayesian model comparison between different covariance functions. We apply our techniques to both synthetic and real data and quantify the speed-up relative to using nested sampling to numerically evaluate model evidences.

Fixed missing references

References in corpus (4)

• MultiNest: an efficient and robust Bayesian inference tool for cosmology and particle physics
• Multimodal nested sampling: an efficient and robust alternative to MCMC methods for astronomical data analysis
• Improving gravitational-wave parameter estimation using Gaussian process regression
• Novel Method for Incorporating Model Uncertainties into Gravitational Wave Parameter Estimates

Cited by in corpus (6)

• Eccentric, nonspinning, inspiral, Gaussian-process merger approximant for the detection and characterization of eccentric binary black hole mergers
• Physics of eccentric binary black hole mergers: A numerical relativity perspective
• Regression methods in waveform modeling: a comparative study
• 2 Fast 2 Fiducial: Gaussian processes for the interpolation and marginalization of waveform error in extreme-mass-ratio-inspiral parameter estimation
• Characterization of numerical relativity waveforms of eccentric binary black hole mergers
• Improving the scalability of Gaussian-process error marginalization in gravitational-wave inference