RL-DEVS : towards discrete event simulation integration into reinfircement learning

Reinforcement learning is a type of machine learning in which an agent learns by repeatedly interacting with an environment. This environment often has to be simulated to allow for an easier learning, hence our proposed integration of reinforcement into a DEVS framework walled RL-DEVS. This article uses an atomic model for the agent and a coupled model for the environement. It borrows existing steps from the reinforcement learning to couple the models in what is called the reinforcement learning loop. This loop already defines the inputs, outputs and possible transitions between both models.