Deep learning with differential Gaussian process flows

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

35 Downloads (Pure)

Abstract

We propose a novel deep learning paradigm of differential flows that learn a stochastic differential equation transformations of inputs prior to a standard classification or regression function. The key property of differential Gaussian processes is the warping of inputs through infinitely deep, but infinitesimal, differential fields, that generalise discrete layers into a dynamical system. We demonstrate state-of-the-art results that exceed the performance of deep Gaussian processes and neural networks
Original languageEnglish
Title of host publicationThe 22nd International Conference on Artificial Intelligence and Statistic
Pages1-15
Number of pages16
Volume89
Publication statusPublished - Apr 2019
MoE publication typeA4 Article in a conference publication
EventInternational Conference on Artificial Intelligence and Statistics - Naha, Japan
Duration: 16 Apr 201918 Apr 2019
Conference number: 22

Publication series

NameProceedings of Machine Learning Research
PublisherPMLR
Volume89
ISSN (Electronic)2640-3498

Conference

ConferenceInternational Conference on Artificial Intelligence and Statistics
Abbreviated titleAISTATS
CountryJapan
CityNaha
Period16/04/201918/04/2019

Keywords

  • gaussian process
  • Bayesian methods

Fingerprint Dive into the research topics of 'Deep learning with differential Gaussian process flows'. Together they form a unique fingerprint.

Cite this