References & Citations
Statistics > Machine Learning
Title: Nested Kriging predictions for datasets with large number of observations
(Submitted on 19 Jul 2016 (v1), last revised 25 Jul 2017 (this version, v3))
Abstract: This work falls within the context of predicting the value of a real function at some input locations given a limited number of observations of this function. The Kriging interpolation technique (or Gaussian process regression) is often considered to tackle such a problem but the method suffers from its computational burden when the number of observation points is large. We introduce in this article nested Kriging predictors which are constructed by aggregating sub-models based on subsets of observation points. This approach is proven to have better theoretical properties than other aggregation methods that can be found in the literature. Contrarily to some other methods it can be shown that the proposed aggregation method is consistent. Finally, the practical interest of the proposed method is illustrated on simulated datasets and on an industrial test case with $10^4$ observations in a 6-dimensional space.
Submission history
From: Nicolas Durrande [view email][v1] Tue, 19 Jul 2016 07:27:02 GMT (412kb,D)
[v2] Thu, 22 Dec 2016 12:06:20 GMT (417kb,D)
[v3] Tue, 25 Jul 2017 12:53:28 GMT (240kb,D)
Link back to: arXiv, form interface, contact.