Sequential sampling of Gaussian process latent variable models

TL;DR

This paper introduces a sequential sampling method for Gaussian process latent variable models, enabling efficient inference on large datasets by exploiting sequential structure, thus overcoming computational limitations of traditional MCMC approaches.

Contribution

It presents a novel approximation technique that allows sequential sampling of latent variables and parameters in Gaussian process models, improving scalability and efficiency.

Findings

Effective in growing-data scenarios

Outperforms naive sampling methods

Scales efficiently with data size

Abstract

We consider the problem of inferring a latent function in a probabilistic model of data. When dependencies of the latent function are specified by a Gaussian process and the data likelihood is complex, efficient computation often involve Markov chain Monte Carlo sampling with limited applicability to large data sets. We extend some of these techniques to scale efficiently when the problem exhibits a sequential structure. We propose an approximation that enables sequential sampling of both latent variables and associated parameters. We demonstrate strong performance in growing-data settings that would otherwise be unfeasible with naive, non-sequential sampling.