Transparency in Sum-Product Network Decompilation

Giannis Papantonis; Vaishak Belle

doi:10.3233/FAIA2304

Transparency in Sum-Product Network Decompilation

Giannis Papantonis, Vaishak Belle

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract / Description of output

Sum-product networks guarantee that conditionals and marginals can be computed efficiently, for a wide range of models, bypassing the hardness of inference. However, this advantage comes at the expense of transparency, since it is unclear how variables inter act in sum-product networks. Due to this, a series of decompilation algorithms transform sum-product networks back to Bayesian networks. In this work, we first study the transparency and causal utility of the resulting Bayesian networks. We then propose a novel decompilation algorithm to address the identified limitations.

Original language	English
Title of host publication	Proceedings of the 26th European Conference on Artificial Intelligence
Publisher	IOS Press
Pages	1827-1834
Number of pages	8
Volume	372
ISBN (Electronic)	9781643684376
ISBN (Print)	9781643684369
DOIs	https://doi.org/10.3233/FAIA2304
Publication status	Published - 1 Oct 2023
Event	26th European Conference on Artificial Intelligence - ICE Kraków Congress Centre, Kraków, Poland Duration: 30 Sept 2023 → 5 Oct 2023 https://ecai2023.eu/

Publication series

Name	Frontiers in Artificial Intelligence and Applications
Publisher	IOS Press
Volume	372
ISSN (Print)	0922-6389
ISSN (Electronic)	1879-8314

Conference

Conference	26th European Conference on Artificial Intelligence
Abbreviated title	ECAI 2023
Country/Territory	Poland
City	Kraków
Period	30/09/23 → 5/10/23
Internet address	https://ecai2023.eu/

Access to Document

10.3233/FAIA2304Licence: Creative Commons: Attribution Non-Commercial (CC-BY-NC)

Transparency in Sum-Product_PAPANTONIS_DOA15072023_AFV_CC BYAccepted author manuscript, 305 KBLicence: Creative Commons: Attribution (CC-BY)
Transparancy in Sum-Product_PAPANTONIS_DOA15072023_VOR_CC BY-NCFinal published version, 316 KBLicence: Creative Commons: Attribution Non-Commercial (CC-BY-NC)

https://ebooks.iospress.nl/volumearticle/64402Licence: Creative Commons: Attribution Non-Commercial (CC-BY-NC)

UKRI Trustworthy Autonomous Systems Node in Governance and Regulation
Ramamoorthy, R., Belle, V., Bundy, A., Jackson, P., Lascarides, A. & Rajan, A.
EPSRC
1/11/20 → 30/04/24
Project: Research

Cite this

@inproceedings{5e1c59fd727f4c81b208e4e1cd747acc,

title = "Transparency in Sum-Product Network Decompilation",

abstract = "Sum-product networks guarantee that conditionals and marginals can be computed efficiently, for a wide range of models, bypassing the hardness of inference. However, this advantage comes at the expense of transparency, since it is unclear how variables inter act in sum-product networks. Due to this, a series of decompilation algorithms transform sum-product networks back to Bayesian networks. In this work, we first study the transparency and causal utility of the resulting Bayesian networks. We then propose a novel decompilation algorithm to address the identified limitations.",

author = "Giannis Papantonis and Vaishak Belle",

year = "2023",

month = oct,

day = "1",

doi = "10.3233/FAIA2304",

language = "English",

isbn = "9781643684369",

volume = "372",

series = "Frontiers in Artificial Intelligence and Applications",

publisher = "IOS Press",

pages = "1827--1834",

booktitle = "Proceedings of the 26th European Conference on Artificial Intelligence",

note = "26th European Conference on Artificial Intelligence, ECAI 2023 ; Conference date: 30-09-2023 Through 05-10-2023",

url = "https://ecai2023.eu/",

}

Papantonis, G & Belle, V 2023, Transparency in Sum-Product Network Decompilation. in Proceedings of the 26th European Conference on Artificial Intelligence. vol. 372, Frontiers in Artificial Intelligence and Applications, vol. 372, IOS Press, pp. 1827-1834, 26th European Conference on Artificial Intelligence, Kraków, Poland, 30/09/23. https://doi.org/10.3233/FAIA2304

TY - GEN

T1 - Transparency in Sum-Product Network Decompilation

AU - Papantonis, Giannis

AU - Belle, Vaishak

PY - 2023/10/1

Y1 - 2023/10/1

N2 - Sum-product networks guarantee that conditionals and marginals can be computed efficiently, for a wide range of models, bypassing the hardness of inference. However, this advantage comes at the expense of transparency, since it is unclear how variables inter act in sum-product networks. Due to this, a series of decompilation algorithms transform sum-product networks back to Bayesian networks. In this work, we first study the transparency and causal utility of the resulting Bayesian networks. We then propose a novel decompilation algorithm to address the identified limitations.

AB - Sum-product networks guarantee that conditionals and marginals can be computed efficiently, for a wide range of models, bypassing the hardness of inference. However, this advantage comes at the expense of transparency, since it is unclear how variables inter act in sum-product networks. Due to this, a series of decompilation algorithms transform sum-product networks back to Bayesian networks. In this work, we first study the transparency and causal utility of the resulting Bayesian networks. We then propose a novel decompilation algorithm to address the identified limitations.

U2 - 10.3233/FAIA2304

DO - 10.3233/FAIA2304

M3 - Conference contribution

SN - 9781643684369

VL - 372

T3 - Frontiers in Artificial Intelligence and Applications

SP - 1827

EP - 1834

BT - Proceedings of the 26th European Conference on Artificial Intelligence

PB - IOS Press

T2 - 26th European Conference on Artificial Intelligence

Y2 - 30 September 2023 through 5 October 2023

ER -

Transparency in Sum-Product Network Decompilation

Abstract / Description of output

Publication series

Conference

Access to Document

Fingerprint

Projects

UKRI Trustworthy Autonomous Systems Node in Governance and Regulation

Cite this