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 CGradientFunction Class Reference Interface for all functions which support gradient update and gradient and output calculation. More... #include <cgradientfunction.h> Inheritance diagram for CGradientFunction: List of all members. Public Member Functions   CGradientFunction (int n_input, int n_output) virtual  ~CGradientFunction () virtual void  getGradient (ColumnVector *input, ColumnVector *outputErrors, CFeatureList *gradientFeatures) virtual void  getFunctionValue (ColumnVector *input, ColumnVector *output)   Interface for calculating the output value. virtual void  getInputDerivation (ColumnVector *input, Matrix *targetVector)   Interface function for calculating the input gradient, only optional and not implemented by all classes. virtual void  getGradientPre (ColumnVector *input, ColumnVector *outputErrors, CFeatureList *gradientFeatures)=0   Interface for calculating the gradient given the input and the outputerror. virtual void  getFunctionValuePre (ColumnVector *input, ColumnVector *output)=0   Interface for calculating the output value. virtual void  getInputDerivationPre (ColumnVector *, Matrix *)   Interface function for calculating the input gradient, only optional and not implemented by all classes. virtual int  getNumInputs ()   Return the dimension of the input. virtual int  getNumOutputs ()   Return the dimension of the output. void  setInputMean (ColumnVector *input_mean) void  setOutputMean (ColumnVector *output_mean) void  setInputStd (ColumnVector *input_std) void  setOutputStd (ColumnVector *output_std) Protected Member Functions virtual void  preprocessInput (ColumnVector *input, ColumnVector *norm_input) virtual void  postprocessOutput (Matrix *norm_output, Matrix *output) Protected Attributes int  num_inputs int  num_outputs ColumnVector *  input_mean ColumnVector *  input_std ColumnVector *  output_mean ColumnVector *  output_std Detailed Description Interface for all functions which support gradient update and gradient and output calculation. The class represents a n-input, m-output function approximator, which depends on numWeights weights. Gradient functions can always calculate the gradient with respect to the weights given an input and the m-dimensional output given an n-dimensional input . This interface extends CGradientUpdateFunction, therefore all subclasses have to implement the same functions as for the super class. Due to the additional functionality of CGradientFunction the subclasses have to implement additionally to the functions from CGradientUpdateFunction following functions: getGradient(ColumnVector *input, ColumnVector *outputErrors, CFeatureList *gradientFeatures) : Calculate the gradient given the input and mthe output error (which is used for backpropagation). Save the gradient in the given feature list, which is supposed to be empty (but its recommended to be sure and clear it before use). getFunctionValue(ColumnVector *input, ColumnVector *output) : Calculate the m-dimensonal output given a n-dimensional input. The vectors are expected to have the correct size! getNumInputs() getNumOutputs() Constructor & Destructor Documentation CGradientFunction::CGradientFunction (  int  n_input, int  n_output )  virtual CGradientFunction::~CGradientFunction (   )  [virtual] Member Function Documentation virtual void CGradientFunction::getFunctionValue (  ColumnVector *  input, ColumnVector *  output )  [virtual]   Interface for calculating the output value. virtual void CGradientFunction::getFunctionValuePre (  ColumnVector *  input, ColumnVector *  output )  [pure virtual]   Interface for calculating the output value. Implemented in CContinuousActionGradientPolicy, and CTorchGradientFunction. virtual void CGradientFunction::getGradient (  ColumnVector *  input, ColumnVector *  outputErrors, CFeatureList *  gradientFeatures )  [virtual] virtual void CGradientFunction::getGradientPre (  ColumnVector *  input, ColumnVector *  outputErrors, CFeatureList *  gradientFeatures )  [pure virtual]   Interface for calculating the gradient given the input and the outputerror. Implemented in CContinuousActionGradientPolicy, and CTorchGradientFunction. virtual void CGradientFunction::getInputDerivation (  ColumnVector *  input, Matrix *  targetVector )  [virtual]   Interface function for calculating the input gradient, only optional and not implemented by all classes. virtual void CGradientFunction::getInputDerivationPre (  ColumnVector *  , Matrix *  )  [inline, virtual]   Interface function for calculating the input gradient, only optional and not implemented by all classes. Reimplemented in CTorchGradientFunction. virtual int CGradientFunction::getNumInputs (   )  [virtual]   Return the dimension of the input. virtual int CGradientFunction::getNumOutputs (   )  [virtual]   Return the dimension of the output. virtual void CGradientFunction::postprocessOutput (  Matrix *  norm_output, Matrix *  output )  [protected, virtual] virtual void CGradientFunction::preprocessInput (  ColumnVector *  input, ColumnVector *  norm_input )  [protected, virtual] void CGradientFunction::setInputMean (  ColumnVector *  input_mean  )  void CGradientFunction::setInputStd (  ColumnVector *  input_std  )  void CGradientFunction::setOutputMean (  ColumnVector *  output_mean  )  void CGradientFunction::setOutputStd (  ColumnVector *  output_std  )  Member Data Documentation ColumnVector* CGradientFunction::input_mean [protected] ColumnVector* CGradientFunction::input_std [protected] int CGradientFunction::num_inputs [protected] int CGradientFunction::num_outputs [protected] ColumnVector* CGradientFunction::output_mean [protected] ColumnVector* CGradientFunction::output_std [protected] The documentation for this class was generated from the following file: cgradientfunction.h