Declarative modelling for Bayesian inference by shallow embedding

By Henrik Nilsson, Thomas A. Nielsen. 2014

6 citations recorded by the source corpus.

A common problem across science and engineering is that aspects of models have to be estimated from observed data. An instance of this familiar to control engineers is system identification. Bayesian inference is a principled way to estimate parameters: exploiting Bayes' theorem, an equational probabilistic model is "inverted", yielding a probability distribution for the unknown parameters given the observations. This paper presents Ebba, a declarative language for probabilistic modelling where models can be used both "forwards" for probabilistic computation and "backwards" for parameter estimation. The novel aspect of Ebba is its implementation: a shallow, arrows-based, embedding. This provides a clear semantical account and ensures that only models that support estimation can be expressed. As arrow-like notions have proved useful in modelling dynamical systems, this might also suggest an approach to an integrated language for modelling dynamical systems and parameter estimation.

Subfields: Artificial Intelligence, Computational Theory and Mathematics, Management Science and Operations Research, Probabilistic & Bayesian Methods

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computational Theory and Mathematics

In Management Science and Operations Research

In Probabilistic & Bayesian Methods

Citations in the corpus

Design and Implementation of Probabilistic Programming Language Anglican (2016, background): “Some languages emphasize expressiveness [6, 7, 9, 10]; others restrict the class of models which can be expressed by the language [11, 13, 23] to facilitate application of efficient inference algorithms.”