Primal Wasserstein Imitation Learning - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2020

Primal Wasserstein Imitation Learning

Robert Dadashi
  • Function : Correspondent author
  • PersonId : 1092948

Connectez-vous pour contacter l'auteur
Léonard Hussenot
  • Function : Author
  • PersonId : 1092830
Matthieu Geist
Olivier Pietquin

Abstract

Imitation Learning (IL) methods seek to match the behavior of an agent with that of an expert. In the present work, we propose a new IL method based on a conceptually simple algorithm: Primal Wasserstein Imitation Learning (PWIL), which ties to the primal form of the Wasserstein distance between the expert and the agent state-action distributions. We present a reward function which is derived offline, as opposed to recent adversarial IL algorithms that learn a reward function through interactions with the environment, and which requires little fine-tuning. We show that we can recover expert behavior on a variety of continuous control tasks of the MuJoCo domain in a sample efficient manner in terms of agent interactions and of expert interactions with the environment. Finally, we show that the behavior of the agent we train matches the behavior of the expert with the Wasserstein distance, rather than the commonly used proxy of performance.
Fichier principal
Vignette du fichier
2006.04678.pdf (6.94 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-03162526 , version 1 (08-03-2021)

Identifiers

Cite

Robert Dadashi, Léonard Hussenot, Matthieu Geist, Olivier Pietquin. Primal Wasserstein Imitation Learning. ICLR 2021 - Ninth International Conference on Learning Representations, May 2021, Vienna / Virtual, Austria. ⟨hal-03162526⟩
94 View
161 Download

Altmetric

Share

Gmail Mastodon Facebook X LinkedIn More