Graph convolutional Gaussian processes
File(s)
Accepted version
Author(s)
Walker, Ian
Glocker, Benjamin
Type
Conference Paper
Abstract
We propose a novel Bayesian nonparametricmethod to learn translation-invariant relationshipson non-Euclidean domains. The resulting graphconvolutional Gaussian processes can be appliedto problems in machine learning for which theinput observations are functions with domains ongeneral graphs. The structure of these models al-lows for high dimensional inputs while retainingexpressibility, as is the case with convolutionalneural networks. We present applications of graphconvolutional Gaussian processes to images andtriangular meshes, demonstrating their versatilityand effectiveness, comparing favorably to existingmethods, despite being relatively simple models.
Date Issued
2019-05-27
Date Acceptance
2019-04-22
Citation
Proceedings of Machine Learning Research, 2019, 97, pp.6495-6504
ISSN
2640-3498
Publisher
PMLR
Start Page
6495
End Page
6504
Journal / Book Title
Proceedings of Machine Learning Research
Volume
97
Copyright Statement
© 2019 The Author(s)
Sponsor
Commission of the European Communities
Grant Number
H2020 - 757173
Source
International Conference on Machine Learning (ICML)
Subjects
cs.LG
cs.CV
stat.ML
Publication Status
Published online
Start Date
2019-06-09
Finish Date
2019-06-15
Coverage Spatial
Long Beach, CA, USA
Date Publish Online
2019-05-27