English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Probabilistic programs with dynamic computation graphs can define measuresover sample spaces with unbounded dimensionality, which constitute programmaticanalogues to Bayesian nonparametrics. Owing to the generality of this modelclass, inference relies on `black-box' Monte Carlo methods that are often notable to take advantage of conditional independence and exchangeability, whichhave historically been the cornerstones of efficient inference. We here seek todevelop a `middle ground' between probabilistic models with fully dynamic andfully static computation graphs. To this end, we introduce a combinator libraryfor the Probabilistic Torch framework. Combinators are functions that acceptmodels and return transformed models. We assume that models are dynamic, butthat model composition is static, in the sense that combinator applicationtakes place prior to evaluating the model on data. Combinators provideprimitives for both model and inference composition. Model combinators take theform of classic functional programming constructs such as map and reduce. Theseconstructs define a computation graph at a coarsened level of representation,in which nodes correspond to models, rather than individual variables.Inference combinators implement operations such as importance resampling andapplication of a transition kernel, which alter the evaluation strategy for amodel whilst preserving proper weighting. Owing to this property, modelsdefined using combinators can be trained using stochastic methods that optimizeeither variational or wake-sleep style objectives. As a validation of thisprinciple, we use combinators to implement black box inference for hiddenMarkov models.

Composing Modeling and Inference Operations with Probabilistic Program Combinators