CFP: Bayesian networks and causal reasoning

[ISHPSS]

 

From: Kbkorb <kbk...@gmail.com>

We are organizing a special journal issue on the topic “Bayesian Networks and Causal Reasoning” in Algorithms, joining us and Judea Pearl, Clark Glymour, David Lagnado and others who’ve agreed to submit papers in this important area of research. We welcome your contribution by 31 Jan 2024 on the use of Bayesian Networks for understanding and explaining causal systems and human causal reasoning, causal discovery (machine learning) or applying causal Bayesian Networks across the sciences.

Algorithms is an open-access journal with a good readership. We hope to have a strong set of publications ranging over topics from the machine learning of causal Bayesian nets, evaluating evidence with causal Bayesian nets,  through modeling human causal reasoning with Bayesian nets. Our formal CFP can be seen here.

We look forward to considering your submission.

P.S. Recently we have made our Causal Attribution Tool (aka CAT) available to the public. This tool allows you to interrogate a causal Bayesian Network to see: updated probabilities given observation evidence (as does every other BN tool, of course); updated probabiltiies given causal interventions plus optional observations (which only a very few tools have allowed before); determine the causal attributions for the effect given those interventions and observations, according to various criteria (which is unique, thus far). Currently, the causal criteria available for this last are limited to FAR (Fraction of Attributable Risk), a version of Patricia Cheng and Clark Glymour’s Causal Power, and an information-theoretic-criterion of our own. Eventually, we hope that a larger range of causal criteria (including those of Halpern & Pearl and Chris Hitchcock) will be made available either by us or by others (the code is open access). Also, we expect to build up a library of causal BNs; currently, there are just a few models there. You can add your own, if you wish to contribute something (just “publish” it before you go!).

-