Multi-Model Probabilistic Programming

TL;DR

This paper introduces an extension to probabilistic programming that enables representation and manipulation of networks of interrelated models, facilitating model search, development tracking, and automation.

Contribution

It provides a formal semantics and algorithms for multi-model probabilistic programs, implemented in Stan, expanding the capabilities of probabilistic modeling.

Findings

Demonstrated automatic model search in Stan

Enabled model development tracking

Proposed applications for model-space navigation

Abstract

Probabilistic programming makes it easy to represent a probabilistic model as a program. Building an individual model, however, is only one step of probabilistic modeling. The broader challenge of probabilistic modeling is in understanding and navigating spaces of alternative models. There is currently no good way to represent these spaces of alternative models, despite their central role. We present an extension of probabilistic programming that lets each program represent a network of interrelated probabilistic models. We give a formal semantics for these multi-model probabilistic programs, a collection of efficient algorithms for network-of-model operations, and an example implementation built on top of the popular probabilistic programming language Stan. This network-of-models representation opens many doors, including search and automation in model-space, tracking and communication of model development, and explicit modeler degrees of freedom to mitigate issues like p-hacking. We demonstrate automatic model search and model development tracking using our Stan implementation, and we propose many more possible applications.