Published On Nov 14, 2018
A DDPG meta-agent for the third Udacity Deep Reinforcement Learning project.
This video presents two results for a trained agent having an average score of 0.5 and another with an average score of 1.0 (at about 2:40 of the video).
One can see, that the movements of the agent with an average score of 1.0 are much crisper, resulting in a more efficient game.
show more