The d-separation criterion in Categorical Probability

The d-separation criterion detects the compatibility of a joint probability distribution with a directed acyclic graph through certain conditional independences. In this work, we study this problem in the context of categorical probability theory by introducing a categorical definition of causal models, a categorical notion of d-separation, and proving an abstract version of the d-separation criterion. This approach has two main benefits. First, categorical d-separation is a very intuitive criterion based on topological connectedness. Second, our results apply both to measure-theoretic probability (with standard Borel spaces) and beyond probability theory, including to deterministic and possibilistic networks. It therefore provides a clean proof of the equivalence of local and global Markov properties with causal compatibility for continuous and mixed random variables as well as deterministic and possibilistic variables.

[1]  W. Liang,et al.  Free gs-Monoidal Categories and Free Markov Categories , 2022, Applied Categorical Structures.

[2]  Michael I. Jordan.  Graphical Models , 2003 .

[3]  Patrick Forr'e,et al.  Transitional Conditional Independence , 2021, 2104.11547.

[4]  E. F. Rischel,et al.  Compositional Abstraction Error and a Category of Causal Models , 2021, UAI.

[5]  C. Heunen,et al.  Categories for Quantum Theory: An Introduction , 2020 .

[6]  Paolo Perrone,et al.  Notes on Category Theory with examples from basic mathematics , 2019, ArXiv.

[7]  T. Fritz.  A synthetic approach to Markov kernels, conditional independence, and theorems on sufficient statistics , 2019, ArXiv.

[8]  Bart Jacobs,et al.  Causal Inference by String Diagram Surgery , 2018, FoSSaCS.

[9]  J. Mooij,et al.  Markov Properties for Graphical Models with Cycles and Latent Variables , 2017, 1710.08775.

[10]  Bart Jacobs,et al.  Disintegration and Bayesian inversion via string diagrams , 2017, Mathematical Structures in Computer Science.

[11]  Fabio Gadducci,et al.  Rewriting modulo symmetric monoidal structure , 2016, 2016 31st Annual ACM/IEEE Symposium on Logic in Computer Science (LICS).

[12]  Judea Pearl,et al.  Causal networks: semantics and expressiveness , 2013, UAI.

[13]  Brendan Fong,et al.  Causal Theories: A Categorical Perspective on Bayesian Networks , 2013, 1301.6201.

[14]  Robert W. Spekkens,et al.  Picturing classical and quantum Bayesian inference , 2011, Synthese.

[15]  B. Coecke.  Quantum picturalism , 2009, 0908.1787.

[16]  John C. Baez,et al.  Physics, Topology, Logic and Computation: A Rosetta Stone , 2009, 0903.0340.

[17]  U. Rieder,et al.  Markov Decision Processes , 2010 .

[18]  Blai Bonet,et al.  Instrumentality Tests Revisited , 2001, UAI.

[19]  Fabio Gadducci,et al.  An Algebraic Presentation of Term Graphs, via GS-Monoidal Categories , 1999, Appl. Categorical Struct..

[20]  Judea Pearl,et al.  On the Testability of Causal Models With Latent and Instrumental Variables , 1995, UAI.

[21]  Axel Poigné,et al.  Basic category theory , 1993, LICS 1993.

[22]  Steffen L. Lauritzen,et al.  Independence properties of directed markov fields , 1990, Networks.

[23]  A. Milosavljevic,et al.  String Diagram Rewriting Modulo Commutative Monoid Structure , 2022, ArXiv.

[24]  E. F. Rischel,et al.  The Category Theory of Causal Models , 2020 .

[25]  Bart Jacobs,et al.  A Formal Semantics of Influence in Bayesian Reasoning , 2017, MFCS.

[26]  J. Pearl.  Causality: Models, Reasoning and Inference , 2000 .

[27]  Fabio Gadducci,et al.  On The Algebraic Approach To Concurrent Term Rewriting , 1996 .

[28]  A. P. Dawid,et al.  Independence properties of directed Markov fields. Networks, 20, 491-505 , 1990 .