Augmented Neural ODEs

  • 2019-04-02 21:50:34
  • Emilien Dupont, Arnaud Doucet, Yee Whye Teh
  • 29

Abstract

We show that Neural Ordinary Differential Equations (ODEs) learnrepresentations that preserve the topology of the input space and prove thatthis implies the existence of functions Neural ODEs cannot represent. Toaddress these limitations, we introduce Augmented Neural ODEs which, inaddition to being more expressive models, are empirically more stable,generalize better and have a lower computational cost than Neural ODEs.

 

Introduction (beta)

None

 

Conclusion (beta)

None