English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Gaussian processes (GPs) provide a framework for Bayesian inference that canoffer principled uncertainty estimates for a large range of problems. Forexample, if we consider regression problems with Gaussian likelihoods, a GPmodel enjoys a posterior in closed form. However, identifying the posterior GPscales cubically with the number of training examples and requires to store allexamples in memory. In order to overcome these obstacles, sparse GPs have beenproposed that approximate the true posterior GP with pseudo-training examples.Importantly, the number of pseudo-training examples is user-defined and enablescontrol over computational and memory complexity. In the general case, sparseGPs do not enjoy closed-form solutions and one has to resort to approximateinference. In this context, a convenient choice for approximate inference isvariational inference (VI), where the problem of Bayesian inference is cast asan optimization problem -- namely, to maximize a lower bound of the logmarginal likelihood. This paves the way for a powerful and versatile framework,where pseudo-training examples are treated as optimization arguments of theapproximate posterior that are jointly identified together with hyperparametersof the generative model (i.e. prior and likelihood). The framework cannaturally handle a wide scope of supervised learning problems, ranging fromregression with heteroscedastic and non-Gaussian likelihoods to classificationproblems with discrete labels, but also problems with multidimensional labels.The purpose of this tutorial is to provide access to the basic matter forreaders without prior knowledge in both GPs and VI. A proper exposition to thesubject enables also access to more recent advances (like importance-weightedVI as well as interdomain, multioutput and deep GPs) that can serve as aninspiration for new research ideas.

A Tutorial on Sparse Gaussian Processes and Variational Inference