Discovery and application of a reinforcement learning algorithm (Q-Learning) on two small projects.
An agent must find a way to heaven by avoiding the hells.
He must therefore learn a strategy because he does not know the environment beforehand.
Two agents play a game of tic-tac-toe.
They have to learn from their actions.
This learning can be stored in a file, and should allow a user to play against the computer.
