Partial Markov Categories [LSC]
Mario Román
Oxford
https://mroman42.github.io/
Date(s) : 12/12/2024 iCal
11h00 - 12h30
Partial Markov categories are a careful blend of Markov categories
(from categorical probability theory) and cartesian restriction
categories (from the algebraic theory of partial computations).
Partial Markov categories are an algebra and syntax for Bayesian
inference. They use a string diagrammatic syntax—with a formal
correspondence to programs—to reason about continuous and discrete
probability, decision problems (Monty Hall, Newcomb’s), the
compositional properties of normalization, and an abstract Bayes’
theorem. We will discuss the construction, theory, and applications of
partial Markov categories; and contextualize them in a line of
research that tries to develop the internal languages of monoidal and
duoidal categories for probability, causality, and process
description. The talk is based on multiple articles, including joint
work with Paweł Sobociński, Elena Di Lavore, Matt Earnshaw, and Bart
Jacobs
Emplacement
I2M Luminy - TPR2, Salle de Séminaire 304-306 (3ème étage)
Catégories