Reinforcement learning as a service
Ivan Čík, Ján Magyar, Marián Mach, Norbert Ferenčík · 2020
This paper describes an application developed in Python that implements reinforcement learning as a service and is ready for deployment on the cloud. The application can be used for solving various types of problems that require reinforcement learning. The application supports the interaction of an agent (e.g. a robot) with an environment in which the agent reaches different states and is able to perform actions (e.g. say a joke). In each state, the agent must select the action most appropriate in the situation, such action selection is provided by our application.