learning GP hyperparameters
(1.6 hours to learn)
Summary
In order to apply Gaussian processes in practice, it is necessary to fit the hyperparameters of the model, such as the lengthscale and variance of a squared-exp kernel. Marginal likelihood is one commonly used criterion for doing so.
Context
This concept has the prerequisites:
- Gaussian process regression (GP regression is the simplest setting in which to discuss hyperparameter learning.)
- Bayesian model comparison (The optimization criterion is based on Bayesian model comparison.)
Core resources (read/watch one of the following)
-Free-
→ Gaussian Processes for Machine Learning
A graduate-level machine learning textbook focusing on Gaussian processes.
Supplemental resources (the following are optional, but you may find them useful)
-Paid-
→ Pattern Recognition and Machine Learning
A textbook for a graduate machine learning course, with a focus on Bayesian methods.
Location:
Section 6.4.3, pages 311-312
See also
-No Additional Notes-