Work

Augmented Neural ODEs

Year: 2019

Type: preprint

Abstract: We show that Neural Ordinary Differential Equations (ODEs) learn representations that preserve the topology of the input space and prove that this implies the existence of functions Neural ODEs cannot represent. To address these limitations, we introduce Augmented Neural ODEs which, in addition to being more expressive models, are empirically more stable, generalize better and have a lower computational cost than Neural ODEs.

Source: arXiv (Cornell University)

DOI: 10.48550/arxiv.1904.01681

Authors: Dupont, Emilien, Doucet, Arnaud, Teh, Yee Whye

Institution: University of Oxford

Language: English

Topic: Neural Networks and Applications

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funders: U.S. Department of Defense, DeepMind, European Commission, Defence Science and Technology Laboratory, Multidisciplinary University Research Initiative, Engineering and Physical Sciences Research Council, Office of Defense Programs

Awards: Semantic Information Pursuit for Multimodal Data Analysis, G5777727104

OpenAlex ID: W2926003515

Cites: 21

Cited by: 285

Related to: 10