Une double approche modulaire de l'apprentissage par renforcement pour des agents intelligents adaptatifs

Abstract : These PhD thesis has been interested in two fields of artificial intelligence: reinforcement learning (RL) on the one hand, and multi-agent systems (MAS) on the other hand. The former allows for the conception of agents (intelligent entities) based on a reinforcement signal which rewards decisions leading to the specified goal, whereas the latter is concerned with the intelligence that can result from the interaction of a group of entities (in the perspective that the whole is more than the sum of its parts). Both these tools suffer from various difficulties. The work we accomplished has shown how these tools can serve each other to answer some of these problems. Thus, agents of a MAS have been conceived through RL, and the architecture of a reinforcement learning agent has been designed as a MAS. Both tools appear to be very complementary, and our global approach of a ``progresssive'' design has proved its efficiency.
Document type :
Theses
Complete list of metadatas

Cited literature [155 references]  Display  Hide  Download

https://hal.univ-lorraine.fr/tel-01748099
Contributor : Thèses Ul <>
Submitted on : Thursday, March 29, 2018 - 11:24:51 AM
Last modification on : Tuesday, April 24, 2018 - 1:53:22 PM
Long-term archiving on : Thursday, September 13, 2018 - 5:49:19 PM

File

SCD_T_2003_0108_BUFFET.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-01748099, version 1

Collections

Citation

Olivier Bernard Henri Buffet. Une double approche modulaire de l'apprentissage par renforcement pour des agents intelligents adaptatifs. Autre [cs.OH]. Université Henri Poincaré - Nancy 1, 2003. Français. ⟨NNT : 2003NAN10108⟩. ⟨tel-01748099⟩

Share

Metrics

Record views

22

Files downloads

15