Monthly Archives: January 2017

An Exponential Family Basis for Probabilistic Programming

Many common distributional families take the form of exponential families.
The functional form of the conditional densities of these families gives them convenient compositional properties, yielding opportunities for powerful reasoning and optimization.

We present a Haskell-based approach for expressing probabilistic models in terms of free arrows, over a basis of exponential families.… Read the rest

Posted in Uncategorized | Comments Off on An Exponential Family Basis for Probabilistic Programming

Encapsulating models and approximate inference programs in probabilistic modules

We introduce the ‘probabilistic module’ interface, which allows encapsulation of complex probabilistic models with latent variables alongside custom stochastic
approximate inference machinery, and provides a platform-agnostic abstraction
barrier separating the model internals from the host probabilistic inference system.
The interface can be seen as a stochastic generalization of a standard simulation
and density interface for probabilistic primitives.… Read the rest

Posted in Uncategorized | Comments Off on Encapsulating models and approximate inference programs in probabilistic modules

Probabilistic Logic Programs: Unifying Program Trace and Possible World Semantics

While statistical relational learning (SRL) and probabilistic programming (PP) both develop rich representation languages and reasoning tools for probabilistic models that naturally deal with a variable number of objects as well as the relationships amongst them, in the past 5 to 8 years SRL and PP have been studied almost in isolation and now have a quite different focus.… Read the rest

Posted in Uncategorized | Comments Off on Probabilistic Logic Programs: Unifying Program Trace and Possible World Semantics

The semantics of Subroutines and Iteration in the Bayesian Programming language ProBT

Authors: R. Laurent, K. Mekhnacha, E. Mazer and P. Bessière

Abstract: Bayesian models are tools of choice when solving problems with incomplete information. Bayesian networks provide a first but limited approach to address such problems. For real world applications, additional semantics is needed to construct more complex models, especially those with repetitive structures or substructures.… Read the rest

Posted in Uncategorized | Comments Off on The semantics of Subroutines and Iteration in the Bayesian Programming language ProBT

GraPPa: Spanning the Expressivity vs. Efficiency Continuum

An important trade-off in the design of PPLs is between the expressiveness of the language of generative models and the power and efficiency of the inference methods. This trade-off between expressiveness and efficiency requires users to choose a different language, and associated toolchain, de- pending on the particular problem they are trying to solve.

Read the rest
Posted in Uncategorized | Comments Off on GraPPa: Spanning the Expressivity vs. Efficiency Continuum

Metropolis-Hastings for Mixtures of Conditional Distributions

Models with embedded conditioning operations — especially with conditioning within conditional branches — are a challenge for Monte-Carlo Markov Chain (MCMC) inference. They are out of scope of the popular Wingate et al. algorithm or many of its variations. Computing the MCMC acceptance ratio in this case has been an open problem.… Read the rest

Posted in Uncategorized | Comments Off on Metropolis-Hastings for Mixtures of Conditional Distributions

Exchangeable Random Processes and Data Abstraction

What is the right notion of module for probabilistic programs?

Purely probabilistic program modules (i.e., those not using internal state) are exchangeable: calls can be commuted and discarded without changing the behaviour.

In general, memory access is neither commutative nor discardable.… Read the rest

Posted in Uncategorized | Comments Off on Exchangeable Random Processes and Data Abstraction

Mathematical structures of probabilistic programming

The design of operational and denotational models for programming languages has historically been a rather post-hoc endeavour, justified by practical concerns such as the validation of program transformations. If one rather views algorithms as “the idiom of modern science” (http://www.cs.princeton.edu/~chazelle/pubs/algorithm.html), then formal semantics is the precondition for models as programs to be sound objects of mathematical scrutiny.… Read the rest

Posted in Uncategorized | Comments Off on Mathematical structures of probabilistic programming