Design and Implementation of Probabilistic Programming Language Anglican

Tolpin, David; Meent, Jan Willem van de; Yang, Hongseok; Wood, Frank

doi:10.48550/arxiv.1608.05263

Cited by 8 publications

(17 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In probabilistic programming, the state space model can be used to model program execution and particle filters are used as one of the general probabilistic programming inference methods. Examples of probabilistic programming languages that use particle filters and SMC for inference include Anglican [6], Biips [7], Birch [8], Figaro [9], LibBi [10], Venture [11], WebPPL [12] and Turing [13].…”

Section: Particle Filtersmentioning

confidence: 99%

Particle Filter with Rejection Control and Unbiased Estimator of the Marginal Likelihood

Kudlicka

Murray

Schön

et al. 2020

ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

We consider the combined use of resampling and partial rejection control in sequential Monte Carlo methods, also known as particle filters. While the variance reducing properties of rejection control are known, there has not been (to the best of our knowledge) any work on unbiased estimation of the marginal likelihood (also known as the model evidence or the normalizing constant) in this type of particle filters. Being able to estimate the marginal likelihood without bias is highly relevant for model comparison, computation of interpretable and reliable confidence intervals, and in exact approximation methods, such as particle Markov chain Monte Carlo. In the paper we present a particle filter with rejection control that enables unbiased estimation of the marginal likelihood.Index Terms-Particle filters, sequential Monte Carlo (SMC), partial rejection control, unbiased estimate of the marginal likelihood

show abstract

Section: Particle Filtersmentioning

confidence: 99%

Particle Filter with Rejection Control and Unbiased Estimator of the Marginal Likelihood

Kudlicka

Murray

Schön

et al. 2020

ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

show abstract

“…This means we can skip re-execution of this part of the program by calling the continuation at α. The implementation in Anglican makes use of this optimization (Tolpin et al, 2016). A second optimization is callsite caching (Ritchie et al, 2016a), which memoizes return values of functions in a manner that accounts for both the argument values and the environment that a function closes over, allowing re-execution in the proposal to be skipped when the arguments and environment are identical.…”

Section: Metropolis-hastingsmentioning

confidence: 99%

An Introduction to Probabilistic Programming

van de Meent,

Paige,

Yang

et al. 2018

Preprint

Self Cite

View full text Add to dashboard Cite

2an interface between program executions and an inference controller.This document closes with a chapter on advanced topics which we believe to be, at the time of writing, interesting directions for probabilistic programming research; directions that point towards a tight integration with deep neural network research and the development of systems for next-generation artificial intelligence applications.We would like to thank the very large number of people who have read through preliminary versions of this manuscript. Comments from the reviewers have been particularly helpful, as well as general interactions with David Blei and Kevin Murphy in particular. Some people we would like to individually thank are, in no particular order,

show abstract

“…are interpretable for non-experts [1]. Further, many state-of-the-art inference methods can be cast in a composable Bayesian way [2] [3] [4].…”

Section: Motivationmentioning

confidence: 99%

“…Once the event arrived the remote system will increase the delay clock for δ in its local time again. With this clock mechanism, we define the resulting joint distribution in factorized way by focusing on the observation and the effect of a delay on it: P (sprinkler|humidity, δ) = P (sprinkler|humidity * ) P (humidity * |humidity, δ) delayed observation (1) We denote the modified observation with humidity * , which can be seen as the traditional conditional distribution.…”

Section: Proposal: Delay-dependent Inferencementioning

confidence: 99%

Techreport: Time-sensitive probabilistic inference for the edge

Weilbach,

Bieniusa

2017

Preprint

View full text Add to dashboard Cite

In recent years the two trends of edge computing and artificial intelligence became both crucial for information processing infrastructures. While the centralized analysis of massive amounts of data seems to be at odds with computation on the outer edge of distributed systems, we explore the properties of eventually consistent systems and statistics to identify sound formalisms for probabilistic inference on the edge. In particular we treat time itself as a random variable that we incorporate into statistical models through probabilistic programming.

show abstract

Design and Implementation of Probabilistic Programming Language Anglican

Cited by 8 publications

References 0 publications

Particle Filter with Rejection Control and Unbiased Estimator of the Marginal Likelihood

Particle Filter with Rejection Control and Unbiased Estimator of the Marginal Likelihood

An Introduction to Probabilistic Programming

Techreport: Time-sensitive probabilistic inference for the edge

Contact Info

Product

Resources

About