Skip to Main content Skip to Navigation

Apprentissage par renforcement développemental

Abstract : Reinforcement learning allows an agent to learn a behavior that has never been previously defined by humans. The agent discovers the environment and the different consequences of its actions through its interaction: it learns from its own experience, without having pre-established knowledge of the goals or effects of its actions. This thesis tackles how deep learning can help reinforcement learning to handle continuous spaces and environments with many degrees of freedom in order to solve problems closer to reality. Indeed, neural networks have a good scalability and representativeness. They make possible to approximate functions on continuous spaces and allow a developmental approach, because they require little a priori knowledge on the domain. We seek to reduce the amount of necessary interaction of the agent to achieve acceptable behavior. To do so, we proposed the Neural Fitted Actor-Critic framework that defines several data efficient actor-critic algorithms. We examine how the agent can fully exploit the transitions generated by previous behaviors by integrating off-policy data into the proposed framework. Finally, we study how the agent can learn faster by taking advantage of the development of his body, in particular, by proceeding with a gradual increase in the dimensionality of its sensorimotor space
Complete list of metadatas

Cited literature [149 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Thursday, March 15, 2018 - 3:36:10 PM
Last modification on : Monday, May 4, 2020 - 11:38:44 AM
Long-term archiving on: : Monday, September 10, 2018 - 9:37:21 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01735202, version 1


Matthieu Zimmer. Apprentissage par renforcement développemental. Intelligence artificielle [cs.AI]. Université de Lorraine, 2018. Français. ⟨NNT : 2018LORR0008⟩. ⟨tel-01735202⟩



Record views


Files downloads