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 CSemiMDPListener Class Reference Interface for all SemiMDP Listeners. More... #include <cagentlistener.h> Inheritance diagram for CSemiMDPListener: List of all members. Public Member Functions   CSemiMDPListener () virtual void  nextStep (CStateCollection *, CAction *, CStateCollection *)   sends the Listener the S-A-S tuple from a new step virtual void  intermediateStep (CStateCollection *, CAction *, CStateCollection *)   sends the Listener the S-A-S tuple from a indermediate step virtual void  newEpisode ()   tells the Listener that a new Episode has startet. Public Attributes bool  enabled Detailed Description Interface for all SemiMDP Listeners. This class is the base class of all Learning and Logging objects. If the listeners get added to a CSemiMarkovDecisionProcess the listener gets informed about all Steps from the SMDP and wether to start a new Episode. There are 3 different kind of events which can be sent to the Listener: nextStep(CStateCollection *oldState, CAction *action, CStateCollection *nextState). The Listener gets S-A-S Tuple of the current step. Usually this is the only data a learning algorithm needs (including the reward which ca be calculated from the S-A-S tuple). intermediateStep(CStateCollection *oldState, CAction *action, CStateCollection *nextState). Only needed for Hierarchical Reinforcement Learning. The Listener gets the same data as in nextStep, but instead of a double step, the S-A-S tuple comes from an intermediate step from an extended Action (see CExtendedAction for Details about intermediate steps). The difference is made because intermediate steps has to be treated differently in some cases (e.g. ETraces) newEpisode(). Indicates that a new Episode has startet. One or more of this event-functions should be implemented by all subclasses. The class is also subclass of CParameterObject, so the paramters of the listeners can be set through that interface See also: CSemiMDPSender CSemiMarkovDecisionProcess CParamterObject Constructor & Destructor Documentation CSemiMDPListener::CSemiMDPListener (   )  [inline] Member Function Documentation virtual void CSemiMDPListener::intermediateStep (  CStateCollection *  , CAction *  , CStateCollection *  )  [inline, virtual]   sends the Listener the S-A-S tuple from a indermediate step Reimplemented in CSemiMDPRewardListener, CHierarchicalController, and CAbstractFeatureStochasticEstimatedModel. virtual void CSemiMDPListener::newEpisode (   )  [inline, virtual]   tells the Listener that a new Episode has startet. Reimplemented in CActorFromQFunction, CActorFromActionValue, CActorFromContinuousActionGradientPolicy, CHierarchicalSemiMarkovDecisionProcess, CDeterministicController, CAdaptiveParameterFromNEpisodesCalculator, CAgentLogger, CEpisodeOutput, CEpisodeMatlabOutput, CContinuousActionRandomPolicy, CEpisode, CBatchEpisodeUpdate, CAverageRewardCalculator, CRewardPerEpisodeCalculator, CValueCalculator, CAverageRewardSameStateCalculator, CValueSameStateCalculator, CPolicyGreedynessEvaluator, CVisitStateCounter, CVisitStateActionCounter, CVisitStateActionEstimator, CSelectiveExplorationCalculator, CGridWorldController, CSubGoalOutput, CHierarchicalController, CBatchStepUpdate, CLSTDLambda, CPEGASUSAnalyticalPolicyGradientCalculator, CPendulumUpTimeCalculator, CGPOMDPGradientCalculator, CPolicyGradientWeightDecayListener, CAverageReinforcementBaseLineCalculator, CREINFORCELearner, CRewardEpisode, CRewardLogger, CGraphDebugger, CTDLearner, CTDResidualLearner, CVPolicyLearner, CVPolicyLearner, CVFunctionLearner, and CVFunctionResidualLearner. virtual void CSemiMDPListener::nextStep (  CStateCollection *  , CAction *  , CStateCollection *  )  [inline, virtual]   sends the Listener the S-A-S tuple from a new step Reimplemented in CDeterministicController, CSemiMDPRewardListener, CAdaptiveParameterFromNStepsCalculator, CAgentLogger, CStateOutput, CActionOutput, CContinuousActionRandomPolicy, CEpisode, CVisitStateCounter, CVisitStateActionCounter, CVisitStateActionEstimator, CSelectiveExplorationCalculator, CGridWorldController, CSubGoalOutput, CHierarchicalController, CBatchStepUpdate, CPEGASUSAnalyticalPolicyGradientCalculator, CPendulumUpTimeCalculator, CPrioritizedSweeping, CContinuousStateList, CSamplingBasedTransitionModel, CAbstractFeatureStochasticEstimatedModel, CDiscreteStochasticEstimatedModel, CFeatureStochasticEstimatedModel, and CAdaptiveParameterFromValueCalculator. Member Data Documentation bool CSemiMDPListener::enabled The documentation for this class was generated from the following file: cagentlistener.h