Monthly Archives: December 2017

Constructive probabilistic semantics with non-spatial locales

Ideally, a probabilistic programming language should admit a computable semantics. But languages often provide operators that denote uncomputable functions, such as comparison of real numbers. While the use of these uncomputable operators may result in uncomputable programs, a programmer can …

Posted in Uncategorized | Comments Off on Constructive probabilistic semantics with non-spatial locales

TensorFlow Distributions

The TensorFlow Distributions library implements a vision of probability theory adapted to the modern deep-learning paradigm of end-to-end differentiable computation. Building on two basic abstractions, it offers flexible building blocks for probabilistic computation. Distributions provide fast, numerically stable methods for …

Posted in Uncategorized | Comments Off on TensorFlow Distributions

Stable, measurable functions and probabilistic programs

In 2014, we proved that the denotational model PCoh of probabilistic coherence spaces and entire functions is fully abstract with respect to the functional “ideal” language PCF enriched with discrete probabilistic distributions (such as a term “coin” evaluating to …

Posted in Uncategorized | Comments Off on Stable, measurable functions and probabilistic programs

Using probabilistic programs as proposals

We show how custom Monte Carlo proposal distributions can be expressed as samplers written in probabilistic programming languages. We call these probabilistic programs proposal programs. Proposal programs allow the inference practitioner to naturally express their knowledge about a target …

Posted in Uncategorized | Comments Off on Using probabilistic programs as proposals

SlicStan: Improving Probabilistic Programming using Information Flow Analysis

Probabilistic programming languages provide a concise and abstract way to specify probabilistic models, while hiding away the underlying inference algorithm. However, those languages are often either not efficient enough to use in practice, or restrict the range of supported models …

Posted in Uncategorized | Comments Off on SlicStan: Improving Probabilistic Programming using Information Flow Analysis

Reasoning about Divergences via Span-liftings

We give a semantic framework for formal verifications of continuous probabilistic programming language for the recent relaxations of differential privacy: Renyi differential privacy and zero-Concentrated differential privacy.
These relaxations can be good definitions of data privacy of machine learning mechanisms …

Posted in Uncategorized | Comments Off on Reasoning about Divergences via Span-liftings

Probabilistic Program Inference With Abstractions

Abstraction is a fundamental tool in the analysis and verification of programs. Typically, a program abstraction selectively models particular aspects of the original program while utilizing non-determinism to conservatively account for other behaviors. However, non-deterministic abstractions do not directly apply …

Posted in Uncategorized | Comments Off on Probabilistic Program Inference With Abstractions

Probabilistic Models for Assured Position, Navigation, and Timing

Position, Navigation, and Timing (PNT) platforms provide fundamental support for critical infrastructure, ranging from air traffic control, emergency services, telecom, financial markets, personal navigation, power grids, space applications, etc. However, the problem of defining PNT assurance metrics remains open.

We …

Posted in Uncategorized | Tagged , , | Comments Off on Probabilistic Models for Assured Position, Navigation, and Timing