Vanya Markova
Ventseslav Shopov
Institute of Robotics, Bulgarian Academy of Science, Plovdiv
Bulgaria
e-mail: markovavanya@yahoo.com
Abstract:
This article focuses on the application of deep learning techniques in sequential games. The main hypothesis is that sequence to sequence learning approach is applicable and demonstrate better performance than classic reinforcement learning. So autonomous agents trough sequence to sequence approach are capable of discovering good solutions to the problem at hand by learning in dynamic environment.
Key words:
sequence to sequence
sequential games
autonomous agents
Section:
Topics: