winnow.h

基于稀疏网络的精选机器学习模型

// -*-c++-*-
//===========================================================
//=     University of Illinois at Urbana-Champaign          =
//=     Department of Computer Science                      =
//=     Dr. Dan Roth - Cognitive Computation Group          =
//=                                                         =
//=  Project: SNoW                                          =
//=                                                         =
//=   Module: Winnow.h                                      =
//=  Version: 3.2.0                                         =
//=  Authors: Jeff Rosen, Andrew Carlson, Nick Rizzolo      =
//=     Date: xx/xx/98                                      = 
//=                                                         =
//= Comments:                                               =
//===========================================================

#ifndef WINNOW_H__
#define WINNOW_H__

#include "LearningAlgorithm.h"

class GlobalParams;

class Winnow : public LearningAlgorithm
{
  public:
  Winnow( GlobalParams & gp_,
	  double alpha = 2.0,
	  double beta = 0.50,
	  double threshold = 1.0,
	  double defaultWeight = 0.05 );
  

//   Winnow( double alpha, double beta, double threshold, 
// 	 double defaultWeight, GlobalParams & gp_ );

    void UpdateCounts( Target& tar, Example& ex );
    bool PresentExample( Target& tar, Example& ex );
    void PerformPercentageEligibility( Target& tar );
    void TrainingComplete( Target& tar );
    //void Discard( Target& tar );

    // Updates the target based on the example and the Winnow parameters.  The
    // decision on how to update has been moved out of the Update function,
    // thus the third parameter.
    void Update( Target& tar, Example& ex, bool promote );

    void SetTargetActivation( Target& tar, Example& ex );
    double ReturnNormalizedActivation( Target& tar);

    void Show( ostream* out );

    void Read( ifstream& in );
    void Write( ofstream& out );

  private:
    double alpha;
    double beta;
    GlobalParams & globalParams;
};


inline Winnow::Winnow( GlobalParams & gp_, double a, double b, 
		       double t, double d )
  : LearningAlgorithm(t, d), alpha(a), beta(b), globalParams(gp_)
{
}


#endif