Details
| Originalsprache | Englisch |
|---|---|
| Titel des Sammelwerks | Machine Learning and Knowledge Discovery in Databases. Research Track - European Conference, ECML PKDD 2021, Proceedings |
| Herausgeber/-innen | Nuria Oliver, Fernando Pérez-Cruz, Stefan Kramer, Jesse Read, Jose A. Lozano |
| Herausgeber (Verlag) | Springer Science and Business Media Deutschland GmbH |
| Seiten | 682-698 |
| Seitenumfang | 17 |
| ISBN (elektronisch) | 978-3-030-86520-7 |
| ISBN (Print) | 9783030865191 |
| Publikationsstatus | Veröffentlicht - 10 Sept. 2021 |
| Veranstaltung | European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2021 - Bilbao, Spanien Dauer: 13 Sept. 2021 → 17 Sept. 2021 |
Publikationsreihe
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Band | 12976 LNAI |
| ISSN (Print) | 0302-9743 |
| ISSN (elektronisch) | 1611-3349 |
Abstract
Counterfactuals have become a popular technique nowadays for interacting with black-box machine learning models and understanding how to change a particular instance to obtain a desired outcome from the model. However, most existing approaches assume instant materialization of these changes, ignoring that they may require effort and a specific order of application. Recently, methods have been proposed that also consider the order in which actions are applied, leading to the so-called sequential counterfactual generation problem. In this work, we propose a model-agnostic method for sequential counterfactual generation. We formulate the task as a multi-objective optimization problem and present a genetic algorithm approach to find optimal sequences of actions leading to the counterfactuals. Our cost model considers not only the direct effect of an action, but also its consequences. Experimental results show that compared to state-of-the-art, our approach generates less costly solutions, is more efficient and provides the user with a diverse set of solutions to choose from.
ASJC Scopus Sachgebiete
- Mathematik (insg.)
- Theoretische Informatik
- Informatik (insg.)
- Allgemeine Computerwissenschaft
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
Machine Learning and Knowledge Discovery in Databases. Research Track - European Conference, ECML PKDD 2021, Proceedings. Hrsg. / Nuria Oliver; Fernando Pérez-Cruz; Stefan Kramer; Jesse Read; Jose A. Lozano. Springer Science and Business Media Deutschland GmbH, 2021. S. 682-698 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Band 12976 LNAI).
Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung › Peer-Review
}
TY - GEN
T1 - Consequence-Aware Sequential Counterfactual Generation
AU - Naumann, Philip
AU - Ntoutsi, Eirini
PY - 2021/9/10
Y1 - 2021/9/10
N2 - Counterfactuals have become a popular technique nowadays for interacting with black-box machine learning models and understanding how to change a particular instance to obtain a desired outcome from the model. However, most existing approaches assume instant materialization of these changes, ignoring that they may require effort and a specific order of application. Recently, methods have been proposed that also consider the order in which actions are applied, leading to the so-called sequential counterfactual generation problem. In this work, we propose a model-agnostic method for sequential counterfactual generation. We formulate the task as a multi-objective optimization problem and present a genetic algorithm approach to find optimal sequences of actions leading to the counterfactuals. Our cost model considers not only the direct effect of an action, but also its consequences. Experimental results show that compared to state-of-the-art, our approach generates less costly solutions, is more efficient and provides the user with a diverse set of solutions to choose from.
AB - Counterfactuals have become a popular technique nowadays for interacting with black-box machine learning models and understanding how to change a particular instance to obtain a desired outcome from the model. However, most existing approaches assume instant materialization of these changes, ignoring that they may require effort and a specific order of application. Recently, methods have been proposed that also consider the order in which actions are applied, leading to the so-called sequential counterfactual generation problem. In this work, we propose a model-agnostic method for sequential counterfactual generation. We formulate the task as a multi-objective optimization problem and present a genetic algorithm approach to find optimal sequences of actions leading to the counterfactuals. Our cost model considers not only the direct effect of an action, but also its consequences. Experimental results show that compared to state-of-the-art, our approach generates less costly solutions, is more efficient and provides the user with a diverse set of solutions to choose from.
KW - Genetic algorithms
KW - Model-agnostic
KW - Multi-objective optimization
KW - Sequential counterfactuals
UR - http://www.scopus.com/inward/record.url?scp=85115701969&partnerID=8YFLogxK
U2 - 10.48550/arXiv.2104.05592
DO - 10.48550/arXiv.2104.05592
M3 - Conference contribution
AN - SCOPUS:85115701969
SN - 9783030865191
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 682
EP - 698
BT - Machine Learning and Knowledge Discovery in Databases. Research Track - European Conference, ECML PKDD 2021, Proceedings
A2 - Oliver, Nuria
A2 - Pérez-Cruz, Fernando
A2 - Kramer, Stefan
A2 - Read, Jesse
A2 - Lozano, Jose A.
PB - Springer Science and Business Media Deutschland GmbH
T2 - European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2021
Y2 - 13 September 2021 through 17 September 2021
ER -