Details
| Originalsprache | Englisch |
|---|---|
| Seiten (von - bis) | 1495-1515 |
| Seitenumfang | 21 |
| Fachzeitschrift | Machine learning |
| Jahrgang | 107 |
| Ausgabenummer | 8-10 |
| Publikationsstatus | Veröffentlicht - 1 Sept. 2018 |
| Extern publiziert | Ja |
Abstract
Automated machine learning (AutoML) seeks to automatically select, compose, and parametrize machine learning algorithms, so as to achieve optimal performance on a given task (dataset). Although current approaches to AutoML have already produced impressive results, the field is still far from mature, and new techniques are still being developed. In this paper, we present ML-Plan, a new approach to AutoML based on hierarchical planning. To highlight the potential of this approach, we compare ML-Plan to the state-of-the-art frameworks Auto-WEKA, auto-sklearn, and TPOT. In an extensive series of experiments, we show that ML-Plan is highly competitive and often outperforms existing approaches.
ASJC Scopus Sachgebiete
- Informatik (insg.)
- Software
- Informatik (insg.)
- Artificial intelligence
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
in: Machine learning, Jahrgang 107, Nr. 8-10, 01.09.2018, S. 1495-1515.
Publikation: Beitrag in Fachzeitschrift › Artikel › Forschung › Peer-Review
}
TY - JOUR
T1 - ML-Plan
T2 - Automated machine learning via hierarchical planning
AU - Mohr, Felix
AU - Wever, Marcel
AU - Hüllermeier, Eyke
N1 - Publisher Copyright: © 2018, The Author(s).
PY - 2018/9/1
Y1 - 2018/9/1
N2 - Automated machine learning (AutoML) seeks to automatically select, compose, and parametrize machine learning algorithms, so as to achieve optimal performance on a given task (dataset). Although current approaches to AutoML have already produced impressive results, the field is still far from mature, and new techniques are still being developed. In this paper, we present ML-Plan, a new approach to AutoML based on hierarchical planning. To highlight the potential of this approach, we compare ML-Plan to the state-of-the-art frameworks Auto-WEKA, auto-sklearn, and TPOT. In an extensive series of experiments, we show that ML-Plan is highly competitive and often outperforms existing approaches.
AB - Automated machine learning (AutoML) seeks to automatically select, compose, and parametrize machine learning algorithms, so as to achieve optimal performance on a given task (dataset). Although current approaches to AutoML have already produced impressive results, the field is still far from mature, and new techniques are still being developed. In this paper, we present ML-Plan, a new approach to AutoML based on hierarchical planning. To highlight the potential of this approach, we compare ML-Plan to the state-of-the-art frameworks Auto-WEKA, auto-sklearn, and TPOT. In an extensive series of experiments, we show that ML-Plan is highly competitive and often outperforms existing approaches.
KW - Algorithm configuration
KW - Algorithm selection
KW - Automated machine learning
KW - Automated planning
KW - Heuristic search
UR - http://www.scopus.com/inward/record.url?scp=85049584587&partnerID=8YFLogxK
U2 - 10.1007/s10994-018-5735-z
DO - 10.1007/s10994-018-5735-z
M3 - Article
AN - SCOPUS:85049584587
VL - 107
SP - 1495
EP - 1515
JO - Machine learning
JF - Machine learning
SN - 0885-6125
IS - 8-10
ER -