Partial Markov Categories (Di Lavore, Roman, 2024)
Abstract. Partial Markov categories are a synthetic probabilistic inference framework, blending Markov categories with Cartesian Restriction Categories. We will discuss continuous and discrete probability, observations, Bayes’ theorem, normalisation, and both Pearl’s and Jeffrey’s updates in purely abstract categorical terms. The talk is based on joint work with Elena Di Lavore, Bart Jacobs, and Paweł Sobociński (“Partial Markov Categories”, https://arxiv.org/pdf/2502.03477; and “A Simple Formal Language for Probabilistic Decision Theory”, https://arxiv.org/pdf/2410.10643).