Hyperparameter sensitivity revisited

Tomi Silander

Hyperparameter sensitivity revisited

Tomi Silander

Proceedings of The 3rd International Workshop on Advanced Methodologies for Bayesian Networks, PMLR 73:7-7, 2017.

Abstract

The BDeu scoring criterion for learning Bayesian network structures is known to be very sensitive to the equivalent sample size hyper-parameter. Recently some authors have suggested alternative Bayesian scoring criteria that appear to behave better than BDeu. So is the problem solved? We will review the problem and suggested solutions and present empirical assessment of the current situation.

Cite this Paper

BibTeX


@InProceedings{pmlr-v73-silander17a,
  title = 	 {Hyperparameter sensitivity revisited},
  author = 	 {Silander, Tomi},
  booktitle = 	 {Proceedings of The 3rd International Workshop on Advanced Methodologies for Bayesian Networks},
  pages = 	 {7--7},
  year = 	 {2017},
  editor = 	 {Hyttinen, Antti and Suzuki, Joe and Malone, Brandon},
  volume = 	 {73},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {20--22 Sep},
  publisher =    {PMLR},
  pdf = 	 {http://proceedings.mlr.press/v73/silander17a/silander17a.pdf},
  url = 	 {https://proceedings.mlr.press/v73/silander17a.html},
  abstract = 	 {The BDeu scoring criterion for learning Bayesian network structures is known to be very
 sensitive to the equivalent sample size hyper-parameter. Recently some authors have suggested
 alternative Bayesian scoring criteria that appear to behave better than BDeu. So
 is the problem solved? We will review the problem and suggested solutions and present
 empirical assessment of the current situation.
 }
}

Endnote

%0 Conference Paper
%T Hyperparameter sensitivity revisited
%A Tomi Silander
%B Proceedings of The 3rd International Workshop on Advanced Methodologies for Bayesian Networks
%C Proceedings of Machine Learning Research
%D 2017
%E Antti Hyttinen
%E Joe Suzuki
%E Brandon Malone	
%F pmlr-v73-silander17a
%I PMLR
%P 7--7
%U https://proceedings.mlr.press/v73/silander17a.html
%V 73
%X The BDeu scoring criterion for learning Bayesian network structures is known to be very
 sensitive to the equivalent sample size hyper-parameter. Recently some authors have suggested
 alternative Bayesian scoring criteria that appear to behave better than BDeu. So
 is the problem solved? We will review the problem and suggested solutions and present
 empirical assessment of the current situation.

APA


Silander, T.. (2017). Hyperparameter sensitivity revisited. Proceedings of The 3rd International Workshop on Advanced Methodologies for Bayesian Networks, in Proceedings of Machine Learning Research 73:7-7 Available from https://proceedings.mlr.press/v73/silander17a.html.

Related Material

Download PDF