Reinforcement learning agent for learning to play to a modified version of blackjack.
- Monte carlo,
- Sarsa(lambda),
- Sarsa(lambda) + Linear Approx
In order to test the system and see some relevant metrics concerning its performance you can run the file "testing.py"
The code is structured in two main classes
-
Agent: Implements a suitable reinforcement learning algorithm to learn to play blackjack from experience.
-
Environment: Provides feedback to user's actions, by providing a state representation and a reward signal.