Quantifying mismatch in Bayesian optimization

Schulz, E; Speekenbrink, M; Hernández-Lobato, JM; Ghahramani, Z; Gershman, SJ

Item

ITEM ACTIONSEXPORT

Add to Basket

Local TagsRelease HistoryDetailsSummary

Released

Conference Paper

Quantifying mismatch in Bayesian optimization

MPS-Authors

There are no MPG-Authors in the publication available

External Resource

https://bayesopt.github.io/past/2016.html
(Publisher version)

Fulltext (restricted access)

There are currently no full texts shared for your IP range.

Fulltext (public)

There are no public fulltexts stored in PuRe

Supplementary Material (public)

There is no public supplementary material available

Citation

Schulz, E., Speekenbrink, M., Hernández-Lobato, J., Ghahramani, Z., & Gershman, S. (2016). Quantifying mismatch in Bayesian optimization. In NIPS Workshop Bayesian Optimization: Black-box Optimization and Beyond (BayesOpt 2016) (pp. 1890-1895).

Cite as: https://hdl.handle.net/21.11116/0000-0006-B41A-2

Abstract

How does misspecifying prior smoothness assumptions about the target function affect the performance of Bayesian optimization routines? We show that misspecifying smoothness leads to an increase in regret, that this effect gets worse in higher dimensions, and that it remains substantial even if hyper-parameters are optimized.