clarion.system
Interface InterfaceUsesQLearning

All Superinterfaces:
InterfaceHandlesFeedback, InterfaceRuntimeTrainable, InterfaceTrainable
All Known Implementing Classes:
QBPNet

public interface InterfaceUsesQLearning
extends InterfaceRuntimeTrainable

This interface is implemented by classes that use the Q-learning algorithm for learning and updating of match statistics within CLARION. It extends the InterfaceRuntimeTrainable interface.

Usage:

Any class that implements this interface will perform learning and updating of match statistics within CLARION using the q-learning method (see Sun Tutorial, 2003).

Known classes implementing this interface:

Version:
6.0.4
Author:
Nick Wilson

Method Summary
 AbstractOutputChunk getChosenOutput()
          Gets the chosen output from the previous time step.
 double getDiscount()
          Gets the discount factor that is used as part of the Q-learning algorithm (see Sun Tutorial, 2003).
 double getMaxQ()
          Gets the maximum Q-value for the new time step.
 
Methods inherited from interface clarion.system.InterfaceTrainable
backwardPass, getSumSqErrors, setDesiredOutput, setDesiredOutput
 
Methods inherited from interface clarion.system.InterfaceHandlesFeedback
checkMatchCriterion, getFeedback, setFeedback
 

Method Detail

getMaxQ

double getMaxQ()
Gets the maximum Q-value for the new time step. This function is otherwise known as Max(Q(y,b)) in the Q-learning literature.

Returns:
The maximum Q-value for the new time step.

getDiscount

double getDiscount()
Gets the discount factor that is used as part of the Q-learning algorithm (see Sun Tutorial, 2003).

Returns:
The discount factor.

getChosenOutput

AbstractOutputChunk getChosenOutput()
Gets the chosen output from the previous time step.

Returns:
The chosen output.