CUED Publications database

Expressive priors in Bayesian neural networks: Kernel combinations and periodic functions

Pearce, T and Tsuchida, R and Zaki, M and Brintrup, A and Neely, A (2019) Expressive priors in Bayesian neural networks: Kernel combinations and periodic functions. In: Conference on Uncertainty in Artificial Intelligence (UAI), 2019-7-16 to --.

Full text not available from this repository.

Abstract

© 2019 Association For Uncertainty in Artificial Intelligence (AUAI). All rights reserved. A simple, flexible approach to creating expressive priors in Gaussian process (GP) models makes new kernels from a combination of basic kernels, e.g. summing a periodic and linear kernel can capture seasonal variation with a long term trend. Despite a well-studied link between GPs and Bayesian neural networks (BNNs), the BNN analogue of this has not yet been explored. This paper derives BNN architectures mirroring such kernel combinations. Furthermore, it shows how BNNs can produce periodic kernels, which are often useful in this context. These ideas provide a principled approach to designing BNNs that incorporate prior knowledge about a function. We showcase the practical value of these ideas with illustrative experiments in supervised and reinforcement learning settings.1

Item Type: Conference or Workshop Item (UNSPECIFIED)
Subjects: UNSPECIFIED
Divisions: Div E > Strategy and Policy
Div E > Manufacturing Systems
Depositing User: Cron Job
Date Deposited: 22 May 2019 20:18
Last Modified: 12 Nov 2019 03:57
DOI: