Details
| Originalsprache | Englisch |
|---|---|
| Seiten (von - bis) | 957–979 |
| Seitenumfang | 23 |
| Fachzeitschrift | VLDB Journal |
| Jahrgang | 33 |
| Ausgabenummer | 4 |
| Frühes Online-Datum | 17 Nov. 2023 |
| Publikationsstatus | Veröffentlicht - Juli 2024 |
Abstract
Optimizing a machine learning pipeline for a task at hand requires careful configuration of various hyperparameters, typically supported by an AutoML system that optimizes the hyperparameters for the given training dataset. Yet, depending on the AutoML system’s own second-order meta-configuration, the performance of the AutoML process can vary significantly. Current AutoML systems cannot automatically adapt their own configuration to a specific use case. Further, they cannot compile user-defined application constraints on the effectiveness and efficiency of the pipeline and its generation. In this paper, we propose Caml, which uses meta-learning to automatically adapt its own AutoML parameters, such as the search strategy, the validation strategy, and the search space, for a task at hand. The dynamic AutoML strategy of Caml takes user-defined constraints into account and obtains constraint-satisfying pipelines with high predictive performance.
ASJC Scopus Sachgebiete
- Informatik (insg.)
- Information systems
- Informatik (insg.)
- Hardware und Architektur
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
in: VLDB Journal, Jahrgang 33, Nr. 4, 07.2024, S. 957–979.
Publikation: Beitrag in Fachzeitschrift › Artikel › Forschung › Peer-Review
}
TY - JOUR
T1 - AutoML in Heavily Constrained Applications
AU - Neutatz, Felix
AU - Lindauer, Marius
AU - Abedjan, Ziawasch
N1 - Publisher Copyright: © The Author(s) 2023.
PY - 2024/7
Y1 - 2024/7
N2 - Optimizing a machine learning pipeline for a task at hand requires careful configuration of various hyperparameters, typically supported by an AutoML system that optimizes the hyperparameters for the given training dataset. Yet, depending on the AutoML system’s own second-order meta-configuration, the performance of the AutoML process can vary significantly. Current AutoML systems cannot automatically adapt their own configuration to a specific use case. Further, they cannot compile user-defined application constraints on the effectiveness and efficiency of the pipeline and its generation. In this paper, we propose Caml, which uses meta-learning to automatically adapt its own AutoML parameters, such as the search strategy, the validation strategy, and the search space, for a task at hand. The dynamic AutoML strategy of Caml takes user-defined constraints into account and obtains constraint-satisfying pipelines with high predictive performance.
AB - Optimizing a machine learning pipeline for a task at hand requires careful configuration of various hyperparameters, typically supported by an AutoML system that optimizes the hyperparameters for the given training dataset. Yet, depending on the AutoML system’s own second-order meta-configuration, the performance of the AutoML process can vary significantly. Current AutoML systems cannot automatically adapt their own configuration to a specific use case. Further, they cannot compile user-defined application constraints on the effectiveness and efficiency of the pipeline and its generation. In this paper, we propose Caml, which uses meta-learning to automatically adapt its own AutoML parameters, such as the search strategy, the validation strategy, and the search space, for a task at hand. The dynamic AutoML strategy of Caml takes user-defined constraints into account and obtains constraint-satisfying pipelines with high predictive performance.
KW - AutoML
KW - Constraints
KW - Meta-Learning
UR - http://www.scopus.com/inward/record.url?scp=85176733476&partnerID=8YFLogxK
U2 - 10.48550/arXiv.2306.16913
DO - 10.48550/arXiv.2306.16913
M3 - Article
VL - 33
SP - 957
EP - 979
JO - VLDB Journal
JF - VLDB Journal
SN - 1066-8888
IS - 4
ER -