Efficiently sampling functions from Gaussian process posteriors

Authors: James Wilson, Viacheslav Borovitskiy, Alexander Terenin, Peter Mostowsky, Marc Deisenroth

ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In a series of experiments designed to test competing sampling schemes statistical properties and practical ramifications, we demonstrate how decoupled sample paths accurately represent Gaussian process posteriors at a fraction of the usual cost.
Researcher Affiliation Academia 1Imperial College London 2St. Petersburg State University 3St. Petersburg Department of Steklov Mathematical Institute of Russian Academy of Sciences 4University College London.
Pseudocode No The paper does not contain explicit pseudocode blocks or algorithms labeled as such.
Open Source Code Yes 4Code: https://github.com/j-wilson/GPflowSampling
Open Datasets No The paper mentions varying amounts of "training data n" and functions drawn from "known GP priors" but does not specify a named public dataset or provide access information for a training dataset.
Dataset Splits No The paper does not explicitly provide details about training/validation/test dataset splits, only mentioning training and test locations.
Hardware Specification No The paper does not explicitly mention specific hardware specifications such as GPU or CPU models used for experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup Yes Across trials, we varied both the dimensionality d of search spaces X = [0, 1]d and the number of initial basis functions ℓ. We set κ = d, but this choice was not found to greatly influence results. The total number of basis functions allocated to weight-space and decoupled samplers was again matched, so that b = m + ℓ. [...] Results using ℓ {1024, 4096, 16384} initial bases correspond with {light, medium, dark} tones and { , , } markers.