TU Darmstadt / ULB / TUbiblio

Deep Lagrangian Networks: Using physics as model prior for Deep Learning

Lutter, M. ; Ritter, C. ; Peters, J. (2019)
Deep Lagrangian Networks: Using physics as model prior for Deep Learning.
7th International Conference on Learning Representations (ICLR'19). New Orleans, United States (06.-09.05.2019)
Konferenzveröffentlichung, Bibliographie

Kurzbeschreibung (Abstract)

Deep learning has achieved astonishing results on many tasks with large amounts of data and generalization within the proximity of training data. For many important real-world applications, these requirements are unfeasible and additional prior knowledge on the task domain is required to overcome the resulting problems. In particular, learning physics models for model-based control requires robust extrapolation from fewer samples – often collected online in real-time – and model errors may lead to drastic damages of the system. Directly incorporating physical insight has enabled us to obtain a novel deep model learning approach that extrapolates well while requiring fewer samples. As a first example, we propose Deep Lagrangian Networks (DeLaN) as a deep network structure upon which Lagrangian Mechanics have been imposed. DeLaN can learn the equations of motion of a mechanical system (i.e., system dynamics) with a deep network efficiently while ensuring physical plausibility. The resulting DeLaN network performs very well at robot tracking control. The proposed method did not only outperform previous model learning approaches at learning speed but exhibits substantially improved and more robust extrapolation to novel trajectories and learns online in real-time

Typ des Eintrags: Konferenzveröffentlichung
Erschienen: 2019
Autor(en): Lutter, M. ; Ritter, C. ; Peters, J.
Art des Eintrags: Bibliographie
Titel: Deep Lagrangian Networks: Using physics as model prior for Deep Learning
Sprache: Englisch
Publikationsjahr: 2019
Veranstaltungstitel: 7th International Conference on Learning Representations (ICLR'19)
Veranstaltungsort: New Orleans, United States
Veranstaltungsdatum: 06.-09.05.2019
URL / URN: https://openreview.net/forum?id=BklHpjCqKm
Zugehörige Links:
Kurzbeschreibung (Abstract):

Deep learning has achieved astonishing results on many tasks with large amounts of data and generalization within the proximity of training data. For many important real-world applications, these requirements are unfeasible and additional prior knowledge on the task domain is required to overcome the resulting problems. In particular, learning physics models for model-based control requires robust extrapolation from fewer samples – often collected online in real-time – and model errors may lead to drastic damages of the system. Directly incorporating physical insight has enabled us to obtain a novel deep model learning approach that extrapolates well while requiring fewer samples. As a first example, we propose Deep Lagrangian Networks (DeLaN) as a deep network structure upon which Lagrangian Mechanics have been imposed. DeLaN can learn the equations of motion of a mechanical system (i.e., system dynamics) with a deep network efficiently while ensuring physical plausibility. The resulting DeLaN network performs very well at robot tracking control. The proposed method did not only outperform previous model learning approaches at learning speed but exhibits substantially improved and more robust extrapolation to novel trajectories and learns online in real-time

Zusätzliche Informationen:

Erstveröffentlichung

Fachbereich(e)/-gebiet(e): 20 Fachbereich Informatik
20 Fachbereich Informatik > Intelligente Autonome Systeme
Zentrale Einrichtungen
Zentrale Einrichtungen > Centre for Cognitive Science (CCS)
Hinterlegungsdatum: 25 Aug 2020 06:24
Letzte Änderung: 19 Okt 2023 10:54
PPN:
Zugehörige Links:
Export:
Suche nach Titel in: TUfind oder in Google
Frage zum Eintrag Frage zum Eintrag

Optionen (nur für Redakteure)
Redaktionelle Details anzeigen Redaktionelle Details anzeigen