Causal Bayesian networks

Staged tree models, probability trees …Causalan Bayesian networks

November 1, 2020 — September 1, 2020

graphical models
how do science
machine learning

Some kind of alternative graphical formalism for causal independence graphs 🤷?

discrete probability trees, sometimes also called staged tree models. A probability tree is one of the simplest models for representing the causal generative process of a random experiment or stochastic process The semantics are self-explanatory: each node in the tree corresponds to a potential state of the process, and the arrows indicate both the probabilistic transitions and the causal dependencies between them. Unlike CBNs, probability trees can model context-specific causal dependencies. However, probability trees do not explicitly represent conditional independencies, and thus, when a distribution and its causal relations admit a representation both as a probability tree and a CBN, the latter is more compact.

There is a deepmind demonstaration notebook.

1 References

Genewein, McGrath, Déletang, et al. 2020. Algorithms for Causal Reasoning in Probability Trees.” arXiv:2010.12237 [Cs].
Görgen. 2017. An algebraic characterisation of staged trees : their geometry and causal implications.”
Ortega. 2011. Bayesian Causal Induction.” arXiv:1111.0708 [Cs, Stat].
———. 2015. Subjectivity, Bayesianism, and Causality.” Pattern Recognition Letters, Philosophical Aspects of Pattern Recognition,.
Shafer. 1996. The Art of Causal Conjecture.