Learning Neural Event Functions for Ordinary Differential Equations

  • 2020-11-19 18:36:31
  • Ricky T. Q. Chen, Brandon Amos, Maximilian Nickel
  • 0

Abstract

The existing Neural ODE formulation relies on an explicit knowledge of thetermination time. We extend Neural ODEs to implicitly defined terminationcriteria modeled by neural event functions, which can be chained together anddifferentiated through. Neural Event ODEs are capable of modeling discrete(instantaneous) changes in a continuous-time system, without prior knowledge ofwhen these changes should occur or how many such changes should exist. We testour approach in modeling hybrid discrete- and continuous- systems such asswitching dynamical systems and collision in multi-body systems, and we proposesimulation-based training of point processes with applications in discretecontrol.

 

Quick Read (beta)

loading the full paper ...