Measures for the evaluation and comparison of graphical model structures

Gabriel Kronberger; Bogdan Burlacu; Michael Kommenda; Stephan Winkler; Michael Affenzeller

doi:10.1007/978-3-319-74718-7_34

Measures for the evaluation and comparison of graphical model structures

Gabriel Kronberger, Bogdan Burlacu, Michael Kommenda, Stephan Winkler, Michael Affenzeller

Research Center Hagenberg

Research output: Chapter in Book/Report/Conference proceedings › Conference contribution › peer-review

3 Citations (Scopus)

Abstract

Structure learning is the identification of the structure of graphical models based solely on observational data and is NP-hard. An important component of many structure learning algorithms are heuristics or bounds to reduce the size of the search space. We argue that variable relevance rankings that can be easily calculated for many standard regression models can be used to improve the efficiency of structure learning algorithms. In this contribution, we describe measures that can be used to evaluate the quality of variable relevance rankings, especially the well-known normalized discounted cumulative gain (NDCG). We evaluate and compare different regression methods using the proposed measures and a set of linear and non-linear benchmark problems.

Original language	English
Title of host publication	Computer Aided Systems Theory – EUROCAST 2017 - 16th International Conference, Revised Selected Papers
Editors	Roberto Moreno-Diaz, Alexis Quesada-Arencibia, Franz Pichler
Publisher	Springer
Pages	283-290
Number of pages	8
ISBN (Print)	9783319747170
DOIs	https://doi.org/10.1007/978-3-319-74718-7_34
Publication status	Published - 2018
Event	16th International Conference on Computer Aided Systems Theory, EUROCAST 2017 - Las Palmas de Gran Canaria, Spain Duration: 19 Feb 2017 → 24 Feb 2017

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10671 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	16th International Conference on Computer Aided Systems Theory, EUROCAST 2017
Country/Territory	Spain
City	Las Palmas de Gran Canaria
Period	19.02.2017 → 24.02.2017

Keywords

Graphical models
Regression
Structure learning

Access to Document

10.1007/978-3-319-74718-7_34

Cite this

Kronberger, G., Burlacu, B., Kommenda, M., Winkler, S., & Affenzeller, M. (2018). Measures for the evaluation and comparison of graphical model structures. In R. Moreno-Diaz, A. Quesada-Arencibia, & F. Pichler (Eds.), Computer Aided Systems Theory – EUROCAST 2017 - 16th International Conference, Revised Selected Papers (pp. 283-290). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10671 LNCS). Springer. https://doi.org/10.1007/978-3-319-74718-7_34

Kronberger, Gabriel ; Burlacu, Bogdan ; Kommenda, Michael et al. / Measures for the evaluation and comparison of graphical model structures. Computer Aided Systems Theory – EUROCAST 2017 - 16th International Conference, Revised Selected Papers. editor / Roberto Moreno-Diaz ; Alexis Quesada-Arencibia ; Franz Pichler. Springer, 2018. pp. 283-290 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{13037638e57e418f9d04c5959d571632,

title = "Measures for the evaluation and comparison of graphical model structures",

abstract = "Structure learning is the identification of the structure of graphical models based solely on observational data and is NP-hard. An important component of many structure learning algorithms are heuristics or bounds to reduce the size of the search space. We argue that variable relevance rankings that can be easily calculated for many standard regression models can be used to improve the efficiency of structure learning algorithms. In this contribution, we describe measures that can be used to evaluate the quality of variable relevance rankings, especially the well-known normalized discounted cumulative gain (NDCG). We evaluate and compare different regression methods using the proposed measures and a set of linear and non-linear benchmark problems.",

keywords = "Graphical models, Regression, Structure learning",

author = "Gabriel Kronberger and Bogdan Burlacu and Michael Kommenda and Stephan Winkler and Michael Affenzeller",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing AG 2018. Copyright: Copyright 2018 Elsevier B.V., All rights reserved.; 16th International Conference on Computer Aided Systems Theory, EUROCAST 2017 ; Conference date: 19-02-2017 Through 24-02-2017",

year = "2018",

doi = "10.1007/978-3-319-74718-7_34",

language = "English",

isbn = "9783319747170",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "283--290",

editor = "Roberto Moreno-Diaz and Alexis Quesada-Arencibia and Franz Pichler",

booktitle = "Computer Aided Systems Theory – EUROCAST 2017 - 16th International Conference, Revised Selected Papers",

address = "Germany",

}

Kronberger, G , Burlacu, B, Kommenda, M, Winkler, S & Affenzeller, M 2018, Measures for the evaluation and comparison of graphical model structures. in R Moreno-Diaz, A Quesada-Arencibia & F Pichler (eds), Computer Aided Systems Theory – EUROCAST 2017 - 16th International Conference, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10671 LNCS, Springer, pp. 283-290, 16th International Conference on Computer Aided Systems Theory, EUROCAST 2017, Las Palmas de Gran Canaria, Spain, 19.02.2017. https://doi.org/10.1007/978-3-319-74718-7_34

Measures for the evaluation and comparison of graphical model structures. / Kronberger, Gabriel ; Burlacu, Bogdan; Kommenda, Michael et al.
Computer Aided Systems Theory – EUROCAST 2017 - 16th International Conference, Revised Selected Papers. ed. / Roberto Moreno-Diaz; Alexis Quesada-Arencibia; Franz Pichler. Springer, 2018. p. 283-290 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10671 LNCS).

Research output: Chapter in Book/Report/Conference proceedings › Conference contribution › peer-review

TY - GEN

T1 - Measures for the evaluation and comparison of graphical model structures

AU - Kronberger, Gabriel

AU - Burlacu, Bogdan

AU - Kommenda, Michael

AU - Winkler, Stephan

AU - Affenzeller, Michael

PY - 2018

Y1 - 2018

N2 - Structure learning is the identification of the structure of graphical models based solely on observational data and is NP-hard. An important component of many structure learning algorithms are heuristics or bounds to reduce the size of the search space. We argue that variable relevance rankings that can be easily calculated for many standard regression models can be used to improve the efficiency of structure learning algorithms. In this contribution, we describe measures that can be used to evaluate the quality of variable relevance rankings, especially the well-known normalized discounted cumulative gain (NDCG). We evaluate and compare different regression methods using the proposed measures and a set of linear and non-linear benchmark problems.

AB - Structure learning is the identification of the structure of graphical models based solely on observational data and is NP-hard. An important component of many structure learning algorithms are heuristics or bounds to reduce the size of the search space. We argue that variable relevance rankings that can be easily calculated for many standard regression models can be used to improve the efficiency of structure learning algorithms. In this contribution, we describe measures that can be used to evaluate the quality of variable relevance rankings, especially the well-known normalized discounted cumulative gain (NDCG). We evaluate and compare different regression methods using the proposed measures and a set of linear and non-linear benchmark problems.

KW - Graphical models

KW - Regression

KW - Structure learning

UR - http://www.scopus.com/inward/record.url?scp=85041818629&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-74718-7_34

DO - 10.1007/978-3-319-74718-7_34

M3 - Conference contribution

SN - 9783319747170

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 283

EP - 290

BT - Computer Aided Systems Theory – EUROCAST 2017 - 16th International Conference, Revised Selected Papers

A2 - Moreno-Diaz, Roberto

A2 - Quesada-Arencibia, Alexis

A2 - Pichler, Franz

PB - Springer

T2 - 16th International Conference on Computer Aided Systems Theory, EUROCAST 2017

Y2 - 19 February 2017 through 24 February 2017

ER -

Kronberger G , Burlacu B, Kommenda M, Winkler S , Affenzeller M. Measures for the evaluation and comparison of graphical model structures. In Moreno-Diaz R, Quesada-Arencibia A, Pichler F, editors, Computer Aided Systems Theory – EUROCAST 2017 - 16th International Conference, Revised Selected Papers. Springer. 2018. p. 283-290. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-74718-7_34

Measures for the evaluation and comparison of graphical model structures

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this