Day 1, April 14 - Posters
Bayesian optimization with non-Gaussian prior
Start Date
14-4-2020 9:00 AM
End Date
14-4-2020 11:00 AM
Publisher
University of Tennessee at Chattanooga
Place of Publication
Chattanooga (Tenn.)
Abstract
We explore using a Gamma distribution - a maximum entropy distribution - as a prior in a Bayesian optimization problem. We hypothesized that because it modeled strictly positive quantities, it would be a good model for machine learning loss models.
Date
April 2020
Document Type
posters
Language
English
Rights
http://rightsstatement.org/vocab/InC/1.0/
Recommended Citation
Zeglen, Peter, "Bayesian optimization with non-Gaussian prior". ReSEARCH Dialogues Conference proceedings. https://scholar.utc.edu/research-dialogues/2020/day1_posters/35.
Bayesian optimization with non-Gaussian prior
We explore using a Gamma distribution - a maximum entropy distribution - as a prior in a Bayesian optimization problem. We hypothesized that because it modeled strictly positive quantities, it would be a good model for machine learning loss models.