Rodent: Relevance determination in ODE

by   Niklas Heim, et al.

From a set of observed trajectories of a partially observed system, we aim to learn its underlying (physical) process without having to make too many assumptions about the generating model. We start with a very general, over-parameterized ordinary differential equation (ODE) of order N and learn the minimal complexity of the model, by which we mean both the order of the ODE as well as the minimum number of non-zero parameters that are needed to solve the problem. The minimal complexity is found by combining the Variational Auto-Encoder (VAE) with Automatic Relevance Determination (ARD) to the problem of learning the parameters of an ODE which we call Rodent. We show that it is possible to learn not only one specific model for a single process, but a manifold of models representing harmonic signals in general.


page 1

page 2

page 3

page 4


Parameterized Complexity of Multi-winner Determination: More Effort Towards Fixed-Parameter Tractability

We study winner determination for three prevalent k-committee selection ...

RENs: Relevance Encoding Networks

The manifold assumption for high-dimensional data assumes that the data ...

VCE: Variational Convertor-Encoder for One-Shot Generalization

Variational Convertor-Encoder (VCE) converts an image to various styles;...

Optimizing differential equations to fit data and predict outcomes

Many scientific problems focus on observed patterns of change or on how ...

Laplacian Prior Variational Automatic Relevance Determination for Transmission Tomography

In the classic sparsity-driven problems, the fundamental L-1 penalty met...

Sparse Methods for Automatic Relevance Determination

This work considers methods for imposing sparsity in Bayesian regression...