Distributional Gradient Matching for Learning Uncertain Neural Dynamics Models

Author name not available (Why is that?)

Publication date: 22 June 2021

Abstract: Differential equations in general and neural ODEs in particular are an essential technique in continuous-time system identification. While many deterministic learning algorithms have been designed based on numerical integration via the adjoint method, many downstream tasks such as active learning, exploration in reinforcement learning, robust control, or filtering require accurate estimates of predictive uncertainties. In this work, we propose a novel approach towards estimating epistemically uncertain neural ODEs, avoiding the numerical integration bottleneck. Instead of modeling uncertainty in the ODE parameters, we directly model uncertainties in the state space. Our algorithm - distributional gradient matching (DGM) - jointly trains a smoother and a dynamics model and matches their gradients via minimizing a Wasserstein loss. Our experiments show that, compared to traditional approximate inference methods based on numerical integration, our approach is faster to train, faster at predicting previously unseen trajectories, and in the context of neural ODEs, significantly more accurate.

Has companion code repository: https://github.com/lenarttreven/dgm

This page was built for publication: Distributional Gradient Matching for Learning Uncertain Neural Dynamics Models

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6370926)