Scalable Gradients for Stochastic Differential Equations

5 Jan 2020Xuechen LiTing-Kam Leonard WongRicky T. Q. ChenDavid Duvenaud

The adjoint sensitivity method scalably computes gradients of solutions to ordinary differential equations. We generalize this method to stochastic differential equations, allowing time-efficient and constant-memory computation of gradients with high-order adaptive solvers... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK BENCHMARK
Video Prediction CMU Mocap-2 Latent ODE Test Error 5.98 # 2
Video Prediction CMU Mocap-2 Latent SDE Test Error 4.03 # 1

Methods used in the Paper