CORTAMBERT, M. Pierre (2021) Reproductibilité de l'apprentissage par renforcement PRE - Research Project, ENSTA.

[img]
Preview
PDF
466Kb

Abstract

Species conservation management is a major environmental concern. Reinforcement learning algorithms allow today advances in several fields. These algorithms can help to solve environmental problems. However, one of computer science’s biggest challenge in research concerns the reproducibility of reinforcement learning. We are thus interested doing reinforcement learning with a randomized initial conditional and will evaluate the relevance of this method in a reproducibility concern.

Item Type:Thesis (PRE - Research Project)
Subjects:Mathematics and Applications
Earth Sciences and Environmental Engineering
ID Code:8461
Deposited By:pierre Cortambert
Deposited On:16 mars 2022 16:48
Dernière modification:16 mars 2022 16:48

Repository Staff Only: item control page