Unit:
Department of Informatics and TelecommunicationsΠληροφορική
Author:
STAMATELIS GEORGIOS
Supervisors info:
Παναγιώτης Σταματόπουλος, Επίκουρος καθηγητής, Τμήμα Πληροφορικής και Τηλεπικοινωνιών, Εθνικό και Καποδιστριακό Πανεπιστήμιο Αθηνών
Original Title:
The Friv Reinforcement Learning Environment
Translated title:
The Friv Reinforcement Learning Environment
Summary:
Development of an open source environment for evaluating deep reinforcement learning algorithms, and transfer learning between similar tasks based on the popular video game website FRIV. There are only single player games, the only input provided to the agent is through the console, the same input the human player sees.
Main subject category:
Technology - Computer science
Keywords:
Reinforcement Learning, Deep Learning, Convolutional Neural Networks, Computer Vision, Control, Video Games, Sequential Decision Making