Empirische Inferenz Conference Paper 2006

Causal Inference by Choosing Graphs with Most Plausible Markov Kernels

We propose a new inference rule for estimating causal structure that underlies the observed statistical dependencies among n random variables. Our method is based on comparing the conditional distributions of variables given their direct causes (the so-called Markov kernels") for all hypothetical causal directions and choosing the most plausible one. We consider those Markov kernels most plausible, which maximize the (conditional) entropies constrained by their observed first moment (expectation) and second moments (variance and covariance with its direct causes) based on their given domain. In this paper, we discuss our inference rule for causal relationships between two variables in detail, apply it to a real-world temperature data set with known causality and show that our method provides a correct result for the example.

Author(s): Sun, X. and Janzing, D. and Schölkopf, B.
Book Title: Proceedings of the 9th International Symposium on Artificial Intelligence and Mathematics
Journal: Proceedings of the Ninth International Symposium on Artificial Intelligence and Mathematics (AI & Math 2006
Pages: 1-11
Year: 2006
Month: January
Day: 0
Bibtex Type: Conference Paper (inproceedings)
Event Name: ISAIM 2006
Event Place: Fort Lauderdale, FL, USA
Digital: 0
Electronic Archiving: grant_archive
Language: en
Organization: Max-Planck-Gesellschaft
School: Biologische Kybernetik
Links:

BibTex

@inproceedings{5635,
  title = {Causal Inference by Choosing Graphs with Most Plausible Markov Kernels},
  journal = {Proceedings of the Ninth International Symposium on Artificial Intelligence and Mathematics (AI & Math 2006},
  booktitle = {Proceedings of the 9th International Symposium on Artificial Intelligence and Mathematics},
  abstract = {We propose a new inference rule for estimating causal structure that underlies the observed statistical dependencies among n random variables. Our method is based on comparing the conditional distributions of variables given their direct causes (the so-called Markov kernels") for all hypothetical causal directions and choosing the most plausible one. We consider those Markov kernels most plausible, which maximize the (conditional) entropies constrained by their observed first moment (expectation) and second moments (variance and covariance with its direct causes) based on their given domain. In this paper, we discuss our inference rule for causal relationships between two variables in detail, apply it to a real-world temperature data set with known causality and show that our method provides a correct result for the example.},
  pages = {1-11},
  organization = {Max-Planck-Gesellschaft},
  school = {Biologische Kybernetik},
  month = jan,
  year = {2006},
  slug = {5635},
  author = {Sun, X. and Janzing, D. and Sch{\"o}lkopf, B.},
  month_numeric = {1}
}