Partial Markov Categories - Theory and Applications in Bayesian Inference

Explore a technical colloquium talk that delves into partial Markov categories as an algebraic framework for Bayesian inference. Learn how string diagrammatic syntax corresponds to programs and can be applied to reason about continuous and discrete probability, analyze decision problems like Monty Hall and Newcomb's paradox, and understand compositional properties of normalization and abstract Bayes' theorem. Discover the theoretical foundations that combine Markov categories from categorical probability theory with cartesian restriction categories from partial computation theory. Based on collaborative research with Elena Di Lavore and presented at LiCS'23, gain insights into the construction, theoretical underpinnings, and practical applications of partial Markov categories in probabilistic reasoning and decision theory.