Εκπαίδευση ευφυών πρακτόρων μέσω τεχνικών ενισχυτικής μάθησης
Intelligent Agent training using reinforcement learning techniques
View/ Open
Keywords
Ενισχυτική μάθηση ; Ευφυείς πράκτορες ; Unity ; Ml-Agents ; Τεχνητή νοημοσύνηAbstract
Continuous developments in the area of Artificial Intelligence are affecting almost every new technology product, as well as the expectation of the people using those products. This thesis is about the study and use of reinforcement learning as a way of intelligent agent training for more realistic behaviors in computer games and simulations. In particular, a tool is used which can be utilized for making simulations and computer games work as environments in which agents can be trained using deep reinforcement learning. The specific tool used is called ML-Agents and the application development will be done in the Unity game engine.