Information-based Variational Model Reduction of high-dimensional Reaction Networks

TL;DR

This paper introduces scalable, information theory-based variational methods for reducing high-dimensional reaction networks, combining sensitivity analysis and approximate inference to create efficient, low-dimensional models with controlled information loss.

Contribution

It presents a novel combination of information theoretic sensitivity analysis and variational inference for scalable model reduction of complex reaction networks.

Findings

Effective reduction of high-dimensional biochemical networks

Maintains key dynamics with fewer variables and parameters

Controls information loss during reduction

Abstract

In this work we present new scalable, information theory-based variational methods for the efficient model reduction of high-dimensional deterministic and stochastic reaction networks. The proposed methodology combines, (a) information theoretic tools for sensitivity analysis that allow us to identify the proper coarse variables of the reaction network, with (b) variational approximate inference methods for training a best-fit reduced model. This approach takes advantage of both physicochemical modeling and data-based approaches and allows to construct optimal parameterized reduced dynamics in the number of variables, reactions and parameters, while controlling the information loss due to the reduction. We demonstrate the effectiveness of our model reduction method on several complex, high-dimensional chemical reaction networks arising in biochemistry.