Skip to content

Latest commit

 

History

History
15 lines (10 loc) · 807 Bytes

File metadata and controls

15 lines (10 loc) · 807 Bytes

Reinforcement-Learning-in-Blackjack

Implementation of several reinforcement learning algorithms used to play a variation of blackjack

In order to run all the algorithms just run main.py.

This will execute test_all_algorithms() function which runs MC, SARSA and Linear Function Approximation with SARSA with plots showing the results.

Details about other modules:

  • environment.py - contains the step() function and the implementation of the environment
  • rl_algorithms - contains MC, SARSA and Linear Function Approximation
  • plotting.py - contains functions to plot value function, SARSA and LFA results
  • policies.py - place to put the policies, at the moment contains just epsilon greedy policy
  • utilities.py - calculation of mean squared error and conversion of state to feature vector for LFA