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 CQStochasticPolicy Class Reference Stochastic Policy which computes its propabilities from the Q-Values of. More... #include <cpolicies.h> Inheritance diagram for CQStochasticPolicy: List of all members. Public Member Functions   CQStochasticPolicy (CActionSet *actions, CActionDistribution *distribution, CAbstractQFunction *qfunction)   ~CQStochasticPolicy () virtual void  getActionValues (CStateCollection *state, CActionSet *availableActions, double *actionValues, CActionDataSet *actionDataSet=NULL)   Interface function for calculating the action ratings, has to be implemented by the subclasses. virtual void  getActionGradient (CStateCollection *state, CAction *action, CActionData *data, CFeatureList *gradientState)   Interface function for calculating the derivative of an action factor. virtual bool  isDifferentiable () virtual CAbstractQFunction *  getQFunction () Protected Member Functions virtual void  getActionStatistics (CStateCollection *state, CAction *action, CActionStatistics *stat)   returns the action statistics object from the q-function Protected Attributes CAbstractQFunction *  qfunction   QFunction of the policy, needed for action decision. Detailed Description Stochastic Policy which computes its propabilities from the Q-Values of. This stochastic policy calculates its action ratings according to the given Q-Function. The getActionValues function writes the Q-Values in the actionFactors array. The Q-Stochastic Policies also support gradient calculation. The policy is differentiable, if the distribution and the Q-Function are differentiable. The gradient d_actionratings(action) / dw calculated in the function getActionGradient is the same as dQ(s,a)/dw. Constructor & Destructor Documentation CQStochasticPolicy::CQStochasticPolicy (  CActionSet *  actions, CActionDistribution *  distribution, CAbstractQFunction *  qfunction )  CQStochasticPolicy::~CQStochasticPolicy (   )  Member Function Documentation virtual void CQStochasticPolicy::getActionGradient (  CStateCollection *  state, CAction *  action, CActionData *  data, CFeatureList *  gradientState )  [virtual]   Interface function for calculating the derivative of an action factor. The function has to calculate d_actionratings(action)/dw, which is for example dQ(s,a)/dw. Reimplemented from CStochasticPolicy. Reimplemented in CVMStochasticPolicy. virtual void CQStochasticPolicy::getActionStatistics (  CStateCollection *  state, CAction *  action, CActionStatistics *  stat )  [protected, virtual]   returns the action statistics object from the q-function Reimplemented from CStochasticPolicy. virtual void CQStochasticPolicy::getActionValues (  CStateCollection *  state, CActionSet *  availableActions, double *  actionValues, CActionDataSet *  actionDataSet = NULL )  [virtual]   Interface function for calculating the action ratings, has to be implemented by the subclasses. Implements CStochasticPolicy. Reimplemented in CQStochasticExplorationPolicy. virtual CAbstractQFunction* CQStochasticPolicy::getQFunction (   )  [inline, virtual] virtual bool CQStochasticPolicy::isDifferentiable (   )  [virtual]   Reimplemented from CStochasticPolicy. Reimplemented in CVMStochasticPolicy. Member Data Documentation CAbstractQFunction* CQStochasticPolicy::qfunction [protected]   QFunction of the policy, needed for action decision. The documentation for this class was generated from the following file: cpolicies.h