KIT | KIT-Bibliothek | Impressum | Datenschutz

DEEP framework for deep learning

Alic, Andy S.; Antonacci, Marica; Caballer, Miguel; Campos, Isabel; Costantini, Alessandro; David, Mario; Dlugolinsky, Stefan; Donvito, Giacinto; Gomes, Jorge; Hardt, Marcus; Heredia, Ignacio; Hluchy, Ladislav; Ito, Keiichi; Kozlov, Valentin; Lloret, Lara; López García, Alvaro; Marco, Jesus; Matyska, Ludek; Moltó, Germán; Nguyen, Giang; ... mehr

The DEEP-Hybrid-DataCloud project developed a distributed architecture to leverage intensive computing techniques such as needed for deep learning. The DEEP framework applies a hybrid-cloud approach and provides a set of tools and cloud services to cover the whole machine learning cycle: from models creation, training, validation and testing to model serving and sharing. These tools and services in particular include DEEP API for a web access to machine learning models, Pilot testbed with heterogeneous resources to develop and test models, DEEP Marketplace for easy sharing, DevOps approach with Data Science template and CI/CD pipeline to facilitate development, testing, and delivery of machine learning applications. The project recently published its second software release and platform, code named DEEP Rosetta, and provides distributed training facility for machine learning, artificial intelligence and deep learning via EOSC portal.

Zugehörige Institution(en) am KIT Steinbuch Centre for Computing (SCC)
Publikationstyp Poster
Publikationsdatum 23.06.2020
Sprache Englisch
Identifikator KITopen-ID: 1000125204
HGF-Programm 46.12.02 (POF III, LK 01)
Data Activities
Veranstaltung ISC High Performance (2020), online, 22.06.2020 – 25.06.2020
Projektinformation DEEP-HybridDataCloud (EU, H2020, 777435)
Bemerkung zur Veröffentlichung Die Veranstaltung fand wegen der Corona-Pandemie als Online-Event statt
Externe Relationen Siehe auch
Schlagwörter AI/Machine Learning/Deep Learning, Clouds and Distributed Computing, Scientific Software Development
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page