We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.NA

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Fast Gaussian Process Posterior Mean Prediction via Local Cross Validation and Precomputation

Abstract: Gaussian processes (GPs) are Bayesian non-parametric models useful in a myriad of applications. Despite their popularity, the cost of GP predictions (quadratic storage and cubic complexity with respect to the number of training points) remains a hurdle in applying GPs to large data. We present a fast posterior mean prediction algorithm called FastMuyGPs to address this shortcoming. FastMuyGPs is based upon the MuyGPs hyperparameter estimation algorithm and utilizes a combination of leave-one-out cross-validation, batching, nearest neighbors sparsification, and precomputation to provide scalable, fast GP prediction. We demonstrate several benchmarks wherein FastMuyGPs prediction attains superior accuracy and competitive or superior runtime to both deep neural networks and state-of-the-art scalable GP algorithms.
Comments: 9 pages content, 4 figures, 3 tables
Subjects: Machine Learning (cs.LG); Numerical Analysis (math.NA); Machine Learning (stat.ML)
MSC classes: 60G15
ACM classes: G.3
Cite as: arXiv:2205.10879 [cs.LG]
  (or arXiv:2205.10879v1 [cs.LG] for this version)

Submission history

From: Alec Dunton [view email]
[v1] Sun, 22 May 2022 17:38:36 GMT (722kb,D)

Link back to: arXiv, form interface, contact.