Counterfactual explanations and model multiplicity: a relational verification view
File(s)
Accepted version
Author(s)
Leofante, Francesco
Botoeva, Elena
Rajani, Vineet
Type
Conference Paper
Abstract
We study the interplay between counterfactual explanations
and model multiplicity in the context of neural network clas-
sifiers. We show that current explanation methods often pro-
duce counterfactuals whose validity is not preserved under
model multiplicity. We then study the problem of generating
counterfactuals that are guaranteed to be robust to model multiplicity, characterise its complexity and propose an approach to solve this problem using ideas from relational verification.
and model multiplicity in the context of neural network clas-
sifiers. We show that current explanation methods often pro-
duce counterfactuals whose validity is not preserved under
model multiplicity. We then study the problem of generating
counterfactuals that are guaranteed to be robust to model multiplicity, characterise its complexity and propose an approach to solve this problem using ideas from relational verification.
Date Issued
2023-09-02
Date Acceptance
2023-05-19
Citation
Proceedings of the 20th International Conference on Principles of Knowledge Representation and Reasoning, 2023, pp.763-768
ISBN
978-1-956792-02-7
ISSN
2334-1033
Publisher
IJCAI Organization
Start Page
763
End Page
768
Journal / Book Title
Proceedings of the 20th International Conference on Principles of Knowledge Representation and Reasoning
Copyright Statement
© 2023 International Joint Conferences on Artificial Intelligence Organization.
Source
The 20th International Conference on Principles of Knowledge Representation and Reasoning (KR2023)
Publication Status
Published
Start Date
2023-09-02
Finish Date
2023-09-08
Coverage Spatial
Rhodes, Greece
Date Publish Online
2023-09-02