liveBook · Manning

Fit_gpytorch_mll

As explained in Bayesian Optimization in Action

fit_gpytorch_mll is a BoTorch helper function that fits a Gaussian-process model by optimizing a marginal log-likelihood object.

fit_gpytorch_mll is a BoTorch helper that trains a Gaussian-process model by maximizing an observed-data marginal log-likelihood. Initialize a likelihood object, wrap it with the model, and pass the resulting objective to the helper. section 9.2.2

A typical multifidelity training workflow is:

  • Create a SingleTaskMultiFidelityGP with the full training inputs, responses, and the data-fidelity column. section 9.2.2

Training a multifidelity GP

After fitting, evaluate the model on a dense grid to obtain predictive means and confidence regions. The resulting mean can pass through high-fidelity observations while low-fidelity observations still reduce uncertainty.

How Bayesian Optimization in Action covers it

How deep it goes
Explains the likelihood-object setup and call sequence, while deferring optimizer internals and implementation details.
Examples
Uses a SingleTaskMultiFidelityGP, an ExactMarginalLogLikelihood, and posterior predictions with confidence regions.
What it assumes
Assumes a Gaussian-process model, training data, and likelihood object; demonstrates the required marginal-likelihood setup.
What you can do after
Apply it via BoTorch to fit a Gaussian-process model from observed training data.
How it connects
Connects Gaussian processes with marginal likelihood, GPyTorch likelihoods, multifidelity models, numerical precision, and data scaling.
Cover of Bayesian Optimization in Action
liveBookmanning.com
Read Bayesian Optimization in Action on liveBook Get the book at manning.com

Related concepts

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.