Quantifying the probable approximation error of probabilistic inference programs

05/31/2016
by   Marco F. Cusumano-Towner, et al.
0

This paper introduces a new technique for quantifying the approximation error of a broad class of probabilistic inference programs, including ones based on both variational and Monte Carlo approaches. The key idea is to derive a subjective bound on the symmetrized KL divergence between the distribution achieved by an approximate inference program and its true target distribution. The bound's validity (and subjectivity) rests on the accuracy of two auxiliary probabilistic programs: (i) a "reference" inference program that defines a gold standard of accuracy and (ii) a "meta-inference" program that answers the question "what internal random choices did the original approximate inference program probably make given that it produced a particular result?" The paper includes empirical results on inference problems drawn from linear regression, Dirichlet process mixture modeling, HMMs, and Bayesian networks. The experiments show that the technique is robust to the quality of the reference inference program and that it can detect implementation bugs that are not apparent from predictive performance.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
12/07/2016

Measuring the non-asymptotic convergence of sequential Monte Carlo samplers using probabilistic programming

A key limitation of sampling algorithms for approximate inference is tha...
research
05/19/2017

AIDE: An algorithm for measuring the accuracy of probabilistic inference algorithms

Approximate probabilistic inference algorithms are central to many field...
research
07/09/2014

Learning Probabilistic Programs

We develop a technique for generalising from data in which models are sa...
research
06/04/2010

Variational Program Inference

We introduce a framework for representing a variety of interesting probl...
research
06/05/2017

Approximate Program Smoothing Using Mean-Variance Statistics, with Application to Procedural Shader Bandlimiting

This paper introduces a general method to approximate the convolution of...
research
03/01/2021

Meta-Learning an Inference Algorithm for Probabilistic Programs

We present a meta-algorithm for learning a posterior-inference algorithm...
research
10/18/2016

Deep Amortized Inference for Probabilistic Programs

Probabilistic programming languages (PPLs) are a powerful modeling tool,...

Please sign up or login with your details

Forgot password? Click here to reset