Quotient normalized maximum likelihood criterion for learning Bayesian network structures

Tomi Silander; Janne Leppä-Aho; Elias Jääsaari; Teemu Roos

Quotient normalized maximum likelihood criterion for learning Bayesian network structures

Tomi Silander, Janne Leppä-Aho, Elias Jääsaari, Teemu Roos

Department of Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

17 Citations (Scopus)

41 Downloads (Pure)

Abstract

We introduce an information theoretic criterion for Bayesian network structure learning which we call quotient normalized maximum likelihood (qNML). In contrast to the closely related factorized normalized maximum likelihood criterion, qNML satisfies the property of score equivalence. It is also decomposable and completely free of adjustable hyperparameters. For practical computations, we identify a remarkably accurate approximation proposed earlier by Szpankowski and Weinberger. Experiments on both simulated and real data demonstrate that the new criterion leads to parsimonious models with good predictive accuracy.

Original language	English
Title of host publication	International Conference on Artificial Intelligence and Statistics, 9-11 April 2018, Playa Blanca, Lanzarote, Canary Islands
Editors	Amos Storkey, Fernando Perez-Cruz
Publisher	JMLR
Pages	948-957
Number of pages	10
Publication status	Published - 1 Jan 2018
MoE publication type	A4 Conference publication
Event	International Conference on Artificial Intelligence and Statistics - Playa Blanca, Spain Duration: 9 Apr 2018 → 11 Apr 2018 Conference number: 21

Publication series

Name	Proceedings of Machine Learning Research
Publisher	PMLR
Volume	84
ISSN (Electronic)	1938-7228

Conference

Conference	International Conference on Artificial Intelligence and Statistics
Abbreviated title	AISTATS
Country/Territory	Spain
City	Playa Blanca
Period	09/04/2018 → 11/04/2018

Access to Document

silander18aFinal published version, 379 KB

http://proceedings.mlr.press/v84/silander18a.html

Cite this

Silander, T., Leppä-Aho, J., Jääsaari, E., & Roos, T. (2018). Quotient normalized maximum likelihood criterion for learning Bayesian network structures. In A. Storkey, & F. Perez-Cruz (Eds.), International Conference on Artificial Intelligence and Statistics, 9-11 April 2018, Playa Blanca, Lanzarote, Canary Islands (pp. 948-957). (Proceedings of Machine Learning Research; Vol. 84). JMLR. http://proceedings.mlr.press/v84/silander18a.html

Silander, Tomi ; Leppä-Aho, Janne ; Jääsaari, Elias et al. / Quotient normalized maximum likelihood criterion for learning Bayesian network structures. International Conference on Artificial Intelligence and Statistics, 9-11 April 2018, Playa Blanca, Lanzarote, Canary Islands. editor / Amos Storkey ; Fernando Perez-Cruz. JMLR, 2018. pp. 948-957 (Proceedings of Machine Learning Research).

@inproceedings{389fef3af45743acb3a5cfe1820c0491,

title = "Quotient normalized maximum likelihood criterion for learning Bayesian network structures",

abstract = "We introduce an information theoretic criterion for Bayesian network structure learning which we call quotient normalized maximum likelihood (qNML). In contrast to the closely related factorized normalized maximum likelihood criterion, qNML satisfies the property of score equivalence. It is also decomposable and completely free of adjustable hyperparameters. For practical computations, we identify a remarkably accurate approximation proposed earlier by Szpankowski and Weinberger. Experiments on both simulated and real data demonstrate that the new criterion leads to parsimonious models with good predictive accuracy.",

author = "Tomi Silander and Janne Lepp{\"a}-Aho and Elias J{\"a}{\"a}saari and Teemu Roos",

year = "2018",

month = jan,

day = "1",

language = "English",

series = "Proceedings of Machine Learning Research",

publisher = "JMLR",

pages = "948--957",

editor = "Amos Storkey and Fernando Perez-Cruz",

booktitle = "International Conference on Artificial Intelligence and Statistics, 9-11 April 2018, Playa Blanca, Lanzarote, Canary Islands",

address = "United States",

note = "International Conference on Artificial Intelligence and Statistics, AISTATS ; Conference date: 09-04-2018 Through 11-04-2018",

}

Silander, T, Leppä-Aho, J, Jääsaari, E & Roos, T 2018, Quotient normalized maximum likelihood criterion for learning Bayesian network structures. in A Storkey & F Perez-Cruz (eds), International Conference on Artificial Intelligence and Statistics, 9-11 April 2018, Playa Blanca, Lanzarote, Canary Islands. Proceedings of Machine Learning Research, vol. 84, JMLR, pp. 948-957, International Conference on Artificial Intelligence and Statistics, Playa Blanca, Spain, 09/04/2018. <http://proceedings.mlr.press/v84/silander18a.html>

Quotient normalized maximum likelihood criterion for learning Bayesian network structures. / Silander, Tomi; Leppä-Aho, Janne; Jääsaari, Elias et al.
International Conference on Artificial Intelligence and Statistics, 9-11 April 2018, Playa Blanca, Lanzarote, Canary Islands. ed. / Amos Storkey; Fernando Perez-Cruz. JMLR, 2018. p. 948-957 (Proceedings of Machine Learning Research; Vol. 84).

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Quotient normalized maximum likelihood criterion for learning Bayesian network structures

AU - Silander, Tomi

AU - Leppä-Aho, Janne

AU - Jääsaari, Elias

AU - Roos, Teemu

N1 - Conference code: 21

PY - 2018/1/1

Y1 - 2018/1/1

N2 - We introduce an information theoretic criterion for Bayesian network structure learning which we call quotient normalized maximum likelihood (qNML). In contrast to the closely related factorized normalized maximum likelihood criterion, qNML satisfies the property of score equivalence. It is also decomposable and completely free of adjustable hyperparameters. For practical computations, we identify a remarkably accurate approximation proposed earlier by Szpankowski and Weinberger. Experiments on both simulated and real data demonstrate that the new criterion leads to parsimonious models with good predictive accuracy.

AB - We introduce an information theoretic criterion for Bayesian network structure learning which we call quotient normalized maximum likelihood (qNML). In contrast to the closely related factorized normalized maximum likelihood criterion, qNML satisfies the property of score equivalence. It is also decomposable and completely free of adjustable hyperparameters. For practical computations, we identify a remarkably accurate approximation proposed earlier by Szpankowski and Weinberger. Experiments on both simulated and real data demonstrate that the new criterion leads to parsimonious models with good predictive accuracy.

UR - http://www.scopus.com/inward/record.url?scp=85067805917&partnerID=8YFLogxK

UR - http://proceedings.mlr.press/v84/

M3 - Conference article in proceedings

T3 - Proceedings of Machine Learning Research

SP - 948

EP - 957

BT - International Conference on Artificial Intelligence and Statistics, 9-11 April 2018, Playa Blanca, Lanzarote, Canary Islands

A2 - Storkey, Amos

A2 - Perez-Cruz, Fernando

PB - JMLR

T2 - International Conference on Artificial Intelligence and Statistics

Y2 - 9 April 2018 through 11 April 2018

ER -

Quotient normalized maximum likelihood criterion for learning Bayesian network structures

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this