Reinforcement Learning Toolbox 2.0 last updated:

General Documentation Manual Tutorial Class Reference Master Thesis Examples Related Papers Downloads Links News mailto:webmaster Main Page     Class Hierarchy   Compound List   File List   Compound Members   File Members CActorFromQFunction Class Reference Actor who creates his Policy on a Q Function. More... #include <cactorcritic.h> Inheritance diagram for CActorFromQFunction: List of all members. Public Member Functions   CActorFromQFunction (CAbstractQFunction *qFunction)   Creates an Actor using the specified Q-Function to adopt his Policy. virtual  ~CActorFromQFunction () virtual void  receiveError (double critic, CStateCollection *oldState, CAction *Action, CActionData *data=NULL)   Updates the Q-Function. CAbstractQFunction *  getQFunction ()   Returns the used Q-Function. CAbstractQETraces *  getETraces ()   Returns the used ETraces. virtual void  newEpisode ()   resets etraces object Protected Attributes CAbstractQFunction *  qFunction   The Q Function of the actor. CAbstractQETraces *  eTraces   The Etraces used for the QFunction. Detailed Description Actor who creates his Policy on a Q Function. The CActorFromQFunction updates it's Q-Function on the particular state action pair according to the critic he got for that state action pair. Since we are using a Q-Function the actor from Q-Function uses QETraces to boost learning. The policy from the actor is usually a Softmax Policy using the Q-Function, this Policy must be created by the user exclusivly. The Q-Function update for this actor is Q(s,a)_new = Q(s,a)_old + beta * td, where td is the value coming from the critic. See also: CActorFromQFunctionAndPolicy. Constructor & Destructor Documentation CActorFromQFunction::CActorFromQFunction (  CAbstractQFunction *  qFunction  )    Creates an Actor using the specified Q-Function to adopt his Policy. virtual CActorFromQFunction::~CActorFromQFunction (   )  [virtual] Member Function Documentation CAbstractQETraces* CActorFromQFunction::getETraces (   )    Returns the used ETraces. CAbstractQFunction* CActorFromQFunction::getQFunction (   )    Returns the used Q-Function. virtual void CActorFromQFunction::newEpisode (   )  [virtual]   resets etraces object Reimplemented from CSemiMDPListener. virtual void CActorFromQFunction::receiveError (  double  critic, CStateCollection *  oldState, CAction *  Action, CActionData *  data = NULL )  [virtual]   Updates the Q-Function. The actor first updates the Etraces (i.e. mulitply all ETraces with gamma*lambda and then adds the state to the ETraces). Then the Q-Function is updated by the Etraces Object with the value beta * critic. See also: CQETraces Implements CActor. Reimplemented in CActorFromQFunctionAndPolicy. Member Data Documentation CAbstractQETraces* CActorFromQFunction::eTraces [protected]   The Etraces used for the QFunction. CAbstractQFunction* CActorFromQFunction::qFunction [protected]   The Q Function of the actor. The documentation for this class was generated from the following file: cactorcritic.h