ReactiveMP.jl: A Julia Package for Reactive Message Passing-based Bayesian Inference

Abstract

ReactiveMP.jl is a native Julia implementation of reactive message passing-based Bayesian inference in probabilistic graphical models with Factor Graphs. The package does Constrained Bethe Free Energy minimisation and supports both exact and variational Bayesian inference, provides a convenient syntax for model specification and allows for extra factorisation and form constraints specification of the variational family of distributions. In addition, ReactiveMP.jl includes a large range of standard probabilistic models and can easily be extended to custom novel nodes and message update rules. In contrast to non-reactive (imperatively coded) Bayesian inference packages, ReactiveMP.jl scales easily to support inference on a standard laptop for large conjugate models with tens of thousands of variables and millions of nodes.

ReactiveMP.jl: A Julia package for reactive variational Bayesian inference

“…The built-in support for specifying constraints on the BFE enables custom approximation methods in different parts of a factor graph representation of the model. As a result, even with these ongoing developments, our team has tested new sophisticated probabilistic models and published multiple scientific articles in peerreviewed journals and conferences with experimental results that were generated by ReactiveMP.jl [1,5,[8][9][10][11].…”