How deeply does Bayesian Optimization in Action cover variational strategy?

Explains the strategy’s role, constructor inputs, and training placement without detailing its internal prediction equations.

Variational strategy

As explained in Bayesian Optimization in Action

VariationalStrategy is a GPyTorch strategy class that constructs approximate Gaussian process predictions from inducing points and variational parameters.

A variational strategy defines how an approximate Gaussian process produces predictions from inducing points. It updates the variational distribution in response to observations, paralleling a multivariate normal update. section 12.2.3

In GPyTorch, it works with a variational distribution and inducing points to represent a full dataset more compactly. Setting learn_inducing_locations=True lets training move those points toward locations that capture important trends. section 12.2.3

A typical approximate model wires the strategy into the training pipeline:

Create a CholeskyVariationalDistribution using the number of inducing points. section 12.2.3

How Bayesian Optimization in Action covers it

Cover of Bayesian Optimization in Action

liveBook manning.com

Read Bayesian Optimization in Action on liveBook Get the book at manning.com

Related concepts

variational Gaussian processhas part · A variational Gaussian process is a scalable approximation to a Gaussian process whose posterior is represented through variational inference over a finite set of inducing points.

This explainer was generated by Manning's knowledge graph from the text of Bayesian Optimization in Action. The section links point to the parts of the book it draws on.