Variational Gaussian processes for linear inverse problems

Authors: Thibault RANDRIANARISOA, Botond Szabo

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the applicability of the procedure in the numerical analysis of Section 4 and conclude the paper with discussion in Section 5.
Researcher Affiliation Academia Thibault Randrianarisoa Department of Decision Sciences Bocconi University via Roentgen 1, 20136, Milano, MI, Italy thibault.randrianarisoa@unibocconi.it; Botond Szabo Department of Decision Sciences Bocconi University via Roentgen 1, 20136, Milano, MI, Italy botond.szabo@unibocconi.it
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any statement or link indicating the availability of open-source code for the described methodology.
Open Datasets No The paper uses synthetic data generated for the experiments: 'We set the sample size n = 8000, take uniformly distributed covariates on [0, 1), and let j cjj (1+β) sin(jπt), cj = 1 + 0.4 sin( 5πj), j odd, 2.5 + 2 sin( 2πj), j even, for β = 1. The independentobservations are generated as Yi N(Af0(xi), 1), depending on the solution of the forward map Af0 after time T = 10 2.' This is not a publicly available dataset.
Dataset Splits No The paper generates synthetic data for its numerical analysis and does not specify explicit training, validation, or test splits. It directly uses the generated data for evaluation.
Hardware Specification Yes The computations were carried out with a 2,6 GHz Quad-Core Intel Core i7 processor.
Software Dependencies No The paper does not specify any software names with version numbers used for the experiments (e.g., specific programming languages, libraries, or frameworks with their versions).
Experiment Setup Yes We set the sample size n = 8000, take uniformly distributed covariates on [0, 1), and let j cjj (1+β) sin(jπt), cj = 1 + 0.4 sin( 5πj), j odd, 2.5 + 2 sin( 2πj), j even, for β = 1. The independentobservations are generated as Yi N(Af0(xi), 1), depending on the solution of the forward map Af0 after time T = 10 2. We consider the prior with λj = e ξj2 for ξ = 10 1. We consider the population spectral feature method described in (10) and plot the variational approximation of the posterior for m = 6 and m = 3 inducing variables in Figure 1.