In the game theoretical approach, the player takes decisions based on the actual states in a game. A game contains two or more players that can be working cooperatively or compete with each other. Each player is supplied with a policy, where the policy can be used for the decision-making process and also optimized. The decision-making of the player is in the form of the next action.
The three main tasks of the player are as follows,
- Compute new action based on the current state.
- Calculate local utility value based on the previous state and next state.
- Optimize their policy based on the state and the selected action.
In MLPro, you can customize your GT-based policy or import the provided policies in the pool of objects (unavailable at the moment).
.. toctree:: :maxdepth: 1 players/custompolicies