actionvalues – RL Components: ActionValues

class pybrain.rl.learners.valuebased.interface.ActionValueInterface

Interface for different ActionValue modules, like the ActionValueTable or the ActionValueNetwork.

class pybrain.rl.learners.valuebased.interface.ActionValueTable(numStates, numActions, name=None)

Bases: pybrain.structure.modules.table.Table, pybrain.rl.learners.valuebased.interface.ActionValueInterface

A special table that is used for Value Estimation methods in Reinforcement Learning.

getMaxAction(state)

Return the action with the maximal value for the given state.

initialize(value=0.0)

Initialize the whole table with the given value.

class pybrain.rl.learners.valuebased.interface.ActionValueNetwork(dimState, numActions, name=None)

Bases: pybrain.structure.modules.module.Module, pybrain.rl.learners.valuebased.interface.ActionValueInterface

getActionValues(state)

Run forward activation for each of the actions and returns all values.

getMaxAction(state)

Return the action with the maximal value for the given state.