Exploiting Innocuousness in Bayesian Networks

Alexander Motzek; Ralf Möller

doi:10.1007/978-3-319-26350-2_36

Exploiting Innocuousness in Bayesian Networks

Alexander Motzek, Ralf Möller

Institute of Information Systems

Abstract

Boolean combination functions in Bayesian networks, such as noisy-or, are often credited a property stating that inactive dependences (e.g., observed to false) do not ``cause any harm'' and an arc becomes vacuous and could have been left out. However, in classic Bayesian networks we are not able to express this property in local CPDs. By using novel ADBNs, we formalize the innocuousness property in CPDs and extend previous work on context-specific independencies. With an explicit representation of innocuousness in local CPDs, we provide a higher causal accuracy for CPD specifications and open new ways for more efficient and less-restricted reasoning in (A)DBNs.

Original language	English
Title of host publication	AI 2015: Advances in Artificial Intelligence
Editors	Bernhard Pfahringer, Jochen Renz
Number of pages	13
Volume	9457
Place of Publication	Cham
Publisher	Springer International Publishing
Publication date	2015
Pages	411-423
ISBN (Print)	978-3-319-26349-6
ISBN (Electronic)	978-3-319-26350-2
DOIs	https://doi.org/10.1007/978-3-319-26350-2_36
Publication status	Published - 2015
Event	28th Australasian Joint Conference on Artificial Intelligence - Canberra, Australia Duration: 30.11.2015 → 04.12.2015 Conference number: 157849

DFG Research Classification Scheme

409-01 Theoretical Computer Science

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1007/978-3-319-26350-2_36

Cite this

@inproceedings{6f9524e8a1104b8c895926b94cb39f6d,

title = "Exploiting Innocuousness in Bayesian Networks",

abstract = "Boolean combination functions in Bayesian networks, such as noisy-or, are often credited a property stating that inactive dependences (e.g., observed to false) do not ``cause any harm'' and an arc becomes vacuous and could have been left out. However, in classic Bayesian networks we are not able to express this property in local CPDs. By using novel ADBNs, we formalize the innocuousness property in CPDs and extend previous work on context-specific independencies. With an explicit representation of innocuousness in local CPDs, we provide a higher causal accuracy for CPD specifications and open new ways for more efficient and less-restricted reasoning in (A)DBNs.",

author = "Alexander Motzek and Ralf M{\"o}ller",

year = "2015",

doi = "10.1007/978-3-319-26350-2_36",

language = "English",

isbn = "978-3-319-26349-6",

volume = "9457",

series = "Lecture Notes in Computer Science",

publisher = "Springer International Publishing",

pages = "411--423",

editor = "Bernhard Pfahringer and Jochen Renz",

booktitle = "AI 2015: Advances in Artificial Intelligence",

note = "28th Australasian Joint Conference on Artificial Intelligence, AI 2015 ; Conference date: 30-11-2015 Through 04-12-2015",

}

Motzek, A & Möller, R 2015, Exploiting Innocuousness in Bayesian Networks. in B Pfahringer & J Renz (eds), AI 2015: Advances in Artificial Intelligence. vol. 9457, Lecture Notes in Computer Science, vol. 9457, Springer International Publishing, Cham, pp. 411-423, 28th Australasian Joint Conference on Artificial Intelligence, Canberra, Australian Capital Territory, Australia, 30.11.15. https://doi.org/10.1007/978-3-319-26350-2_36

Exploiting Innocuousness in Bayesian Networks. / Motzek, Alexander; Möller, Ralf.
AI 2015: Advances in Artificial Intelligence. ed. / Bernhard Pfahringer; Jochen Renz. Vol. 9457 Cham: Springer International Publishing, 2015. p. 411-423 (Lecture Notes in Computer Science; Vol. 9457).

Research output: Chapters in Books/Reports/Conference Proceedings › Conference contribution › peer-review

TY - GEN

T1 - Exploiting Innocuousness in Bayesian Networks

AU - Motzek, Alexander

AU - Möller, Ralf

N1 - Conference code: 157849

PY - 2015

Y1 - 2015

N2 - Boolean combination functions in Bayesian networks, such as noisy-or, are often credited a property stating that inactive dependences (e.g., observed to false) do not ``cause any harm'' and an arc becomes vacuous and could have been left out. However, in classic Bayesian networks we are not able to express this property in local CPDs. By using novel ADBNs, we formalize the innocuousness property in CPDs and extend previous work on context-specific independencies. With an explicit representation of innocuousness in local CPDs, we provide a higher causal accuracy for CPD specifications and open new ways for more efficient and less-restricted reasoning in (A)DBNs.

AB - Boolean combination functions in Bayesian networks, such as noisy-or, are often credited a property stating that inactive dependences (e.g., observed to false) do not ``cause any harm'' and an arc becomes vacuous and could have been left out. However, in classic Bayesian networks we are not able to express this property in local CPDs. By using novel ADBNs, we formalize the innocuousness property in CPDs and extend previous work on context-specific independencies. With an explicit representation of innocuousness in local CPDs, we provide a higher causal accuracy for CPD specifications and open new ways for more efficient and less-restricted reasoning in (A)DBNs.

UR - https://www.scopus.com/record/display.uri?eid=2-s2.0-84952705899&origin=inward&txGid=589f3f3e655d841d299e6337c354a5e9

U2 - 10.1007/978-3-319-26350-2_36

DO - 10.1007/978-3-319-26350-2_36

M3 - Conference contribution

SN - 978-3-319-26349-6

VL - 9457

T3 - Lecture Notes in Computer Science

SP - 411

EP - 423

BT - AI 2015: Advances in Artificial Intelligence

A2 - Pfahringer, Bernhard

A2 - Renz, Jochen

PB - Springer International Publishing

CY - Cham

T2 - 28th Australasian Joint Conference on Artificial Intelligence

Y2 - 30 November 2015 through 4 December 2015

ER -

Exploiting Innocuousness in Bayesian Networks

Abstract

DFG Research Classification Scheme

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this