Persuasive contrastive explanations for Bayesian networks

Tara Koopman; Silja Renooij

doi:10.1007/978-3-030-86772-0_17

Persuasive contrastive explanations for Bayesian networks

Tara Koopman, Silja Renooij^*

^*Corresponding author for this work

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

Abstract

Explanation in Artificial Intelligence is often focused on providing reasons for why a model under consideration and its outcome are correct. Recently, research in explainable machine learning has initiated a shift in focus on including so-called counterfactual explanations. In this paper we propose to combine both types of explanation in the context of explaining Bayesian networks. To this end we introduce persuasive contrastive explanations that aim to provide an answer to the question Why outcome t instead of t^′? posed by a user. In addition, we propose an algorithm for computing persuasive contrastive explanations. Both our definition of persuasive contrastive explanation and the proposed algorithm can be employed beyond the current scope of Bayesian networks.

Original language	English
Title of host publication	Symbolic and Quantitative Approaches to Reasoning with Uncertainty
Subtitle of host publication	16th European Conference, ECSQARU 2021, Prague, Czech Republic, September 21–24, 2021, Proceedings
Editors	Jirina Vejnarová, Nic Wilson
Publisher	Springer
Pages	229-242
Number of pages	14
Edition	1
ISBN (Electronic)	978-3-030-86772-0
ISBN (Print)	978-3-030-86771-3
DOIs	https://doi.org/10.1007/978-3-030-86772-0_17
Publication status	Published - 22 Sept 2021

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12897
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Bibliographical note

Funding Information:
Acknowledgements. This research was partially funded by the Hybrid Intelligence Center, a 10-year programme funded by the Dutch Ministry of Education, Culture and Science through the Netherlands Organisation for Scientific Research, https://hybrid-intelligence-centre.nl. We would like to thank the anonymous reviewers for their useful and inspiring comments.

Publisher Copyright:
© 2021, Springer Nature Switzerland AG.

Keywords

Bayesian networks
Counterfactuals
Explainable AI

Access to Document

10.1007/978-3-030-86772-0_17

Koopman-Renooij2021_Chapter_PersuasiveContrastiveExplanatiFinal published version, 326 KBLicence: Taverne

Cite this

Koopman, T., & Renooij, S. (2021). Persuasive contrastive explanations for Bayesian networks. In J. Vejnarová, & N. Wilson (Eds.), Symbolic and Quantitative Approaches to Reasoning with Uncertainty: 16th European Conference, ECSQARU 2021, Prague, Czech Republic, September 21–24, 2021, Proceedings (1 ed., pp. 229-242). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12897). Springer. https://doi.org/10.1007/978-3-030-86772-0_17

Koopman, Tara ; Renooij, Silja. / Persuasive contrastive explanations for Bayesian networks. Symbolic and Quantitative Approaches to Reasoning with Uncertainty: 16th European Conference, ECSQARU 2021, Prague, Czech Republic, September 21–24, 2021, Proceedings. editor / Jirina Vejnarová ; Nic Wilson. 1. ed. Springer, 2021. pp. 229-242 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{ccc2e8d59f3b454bb11bd3988aa7308a,

title = "Persuasive contrastive explanations for Bayesian networks",

abstract = "Explanation in Artificial Intelligence is often focused on providing reasons for why a model under consideration and its outcome are correct. Recently, research in explainable machine learning has initiated a shift in focus on including so-called counterfactual explanations. In this paper we propose to combine both types of explanation in the context of explaining Bayesian networks. To this end we introduce persuasive contrastive explanations that aim to provide an answer to the question Why outcome t instead of t′? posed by a user. In addition, we propose an algorithm for computing persuasive contrastive explanations. Both our definition of persuasive contrastive explanation and the proposed algorithm can be employed beyond the current scope of Bayesian networks.",

keywords = "Bayesian networks, Counterfactuals, Explainable AI",

author = "Tara Koopman and Silja Renooij",

note = "Funding Information: Acknowledgements. This research was partially funded by the Hybrid Intelligence Center, a 10-year programme funded by the Dutch Ministry of Education, Culture and Science through the Netherlands Organisation for Scientific Research, https://hybrid-intelligence-centre.nl. We would like to thank the anonymous reviewers for their useful and inspiring comments. Publisher Copyright: {\textcopyright} 2021, Springer Nature Switzerland AG.",

year = "2021",

month = sep,

day = "22",

doi = "10.1007/978-3-030-86772-0\_17",

language = "English",

isbn = "978-3-030-86771-3",

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

publisher = "Springer",

pages = "229--242",

editor = "Jirina Vejnarov{\'a} and Nic Wilson",

booktitle = "Symbolic and Quantitative Approaches to Reasoning with Uncertainty",

edition = "1",

}

Koopman, T & Renooij, S 2021, Persuasive contrastive explanations for Bayesian networks. in J Vejnarová & N Wilson (eds), Symbolic and Quantitative Approaches to Reasoning with Uncertainty: 16th European Conference, ECSQARU 2021, Prague, Czech Republic, September 21–24, 2021, Proceedings. 1 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12897, Springer, pp. 229-242. https://doi.org/10.1007/978-3-030-86772-0_17

Persuasive contrastive explanations for Bayesian networks. / Koopman, Tara; Renooij, Silja.
Symbolic and Quantitative Approaches to Reasoning with Uncertainty: 16th European Conference, ECSQARU 2021, Prague, Czech Republic, September 21–24, 2021, Proceedings. ed. / Jirina Vejnarová; Nic Wilson. 1. ed. Springer, 2021. p. 229-242 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12897).

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

TY - GEN

T1 - Persuasive contrastive explanations for Bayesian networks

AU - Koopman, Tara

AU - Renooij, Silja

N1 - Funding Information: Acknowledgements. This research was partially funded by the Hybrid Intelligence Center, a 10-year programme funded by the Dutch Ministry of Education, Culture and Science through the Netherlands Organisation for Scientific Research, https://hybrid-intelligence-centre.nl. We would like to thank the anonymous reviewers for their useful and inspiring comments. Publisher Copyright: © 2021, Springer Nature Switzerland AG.

PY - 2021/9/22

Y1 - 2021/9/22

N2 - Explanation in Artificial Intelligence is often focused on providing reasons for why a model under consideration and its outcome are correct. Recently, research in explainable machine learning has initiated a shift in focus on including so-called counterfactual explanations. In this paper we propose to combine both types of explanation in the context of explaining Bayesian networks. To this end we introduce persuasive contrastive explanations that aim to provide an answer to the question Why outcome t instead of t′? posed by a user. In addition, we propose an algorithm for computing persuasive contrastive explanations. Both our definition of persuasive contrastive explanation and the proposed algorithm can be employed beyond the current scope of Bayesian networks.

AB - Explanation in Artificial Intelligence is often focused on providing reasons for why a model under consideration and its outcome are correct. Recently, research in explainable machine learning has initiated a shift in focus on including so-called counterfactual explanations. In this paper we propose to combine both types of explanation in the context of explaining Bayesian networks. To this end we introduce persuasive contrastive explanations that aim to provide an answer to the question Why outcome t instead of t′? posed by a user. In addition, we propose an algorithm for computing persuasive contrastive explanations. Both our definition of persuasive contrastive explanation and the proposed algorithm can be employed beyond the current scope of Bayesian networks.

KW - Bayesian networks

KW - Counterfactuals

KW - Explainable AI

U2 - 10.1007/978-3-030-86772-0_17

DO - 10.1007/978-3-030-86772-0_17

M3 - Conference contribution

SN - 978-3-030-86771-3

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

SP - 229

EP - 242

BT - Symbolic and Quantitative Approaches to Reasoning with Uncertainty

A2 - Vejnarová, Jirina

A2 - Wilson, Nic

PB - Springer

ER -

Koopman T, Renooij S. Persuasive contrastive explanations for Bayesian networks. In Vejnarová J, Wilson N, editors, Symbolic and Quantitative Approaches to Reasoning with Uncertainty: 16th European Conference, ECSQARU 2021, Prague, Czech Republic, September 21–24, 2021, Proceedings. 1 ed. Springer. 2021. p. 229-242. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-86772-0_17

Persuasive contrastive explanations for Bayesian networks

Abstract

Publication series

Bibliographical note

Keywords

Access to Document

Fingerprint

Cite this