sorteddata.h

MultiBoost 是c++实现的多类adaboost酸法。与传统的adaboost算法主要解决二类分类问题不同

/*
* This file is part of MultiBoost, a multi-class 
* AdaBoost learner/classifier
*
* Copyright (C) 2005-2006 Norman Casagrande
* For informations write to [e-mail]
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
*
*/

/**
* \file SortedData.h Input data which has the column sorted.
*/

#ifndef __SORTED_DATA_H
#define __SORTED_DATA_H

#include "IO/InputData.h"

#include <vector>
#include <utility> // for pair

using namespace std;

namespace MultiBoost {

//////////////////////////////////////////////////////////////////////////////////////////////
//////////////////////////////////////////////////////////////////////////////////////////////

/**
* Overloading of the InputData class to support sorting of the column.
* This is particularly useful for stump-based learner, because
* they work column-by-column (dimension-by-dimension), looking for a threshold
* that minimizes the error, and sorting the data it's mandatory.
* The connection between this class and the weak learner that implements
* decision stump, is done with the overriding of method BaseLearner::createInputData()
* which will return the desired InputData type (and which might depend on
* the arguments of the command line too).
* @see BaseLearner::createInputData()
* @see StumpLearner::createInputData()
* @date 21/11/2005
*/
class SortedData : public InputData
{
public:

   /**
   * The destructor. Must be declared (virtual) for the proper destruction of 
   * the object.
   */
   virtual ~SortedData() {}

   /**
   * Overloading of the load function to support sorting.
   * @param fileName The name of the file to be loaded.
   * @param inputType The type of input.
   * @param verboseLevel The level of verbosity.
   * @see InputData::load()
   * @date 21/11/2005
   * @todo Erase the original memory once sorted?
   */
   virtual void load(const string& fileName, eInputType inputType = IT_TRAIN, int verboseLevel = 1);

   /**
   * Get the first element of the (sorted) column of the data.
   * @param colIdx The column index
   * @return The iterator to the first element of the column
   * @date 10/11/2005
   */
   virtual vpIterator getSortedBegin(int colIdx) { return _sortedData[colIdx].begin(); }

   /**
   * Get the last element of the (sorted) column of the data.
   * @param colIdx The column index
   * @return The iterator to the last element (end()) of the column
   * @remark It is the end() iterator, so it does not point to anything!
   * @date 10/11/2005
   */
   virtual vpIterator getSortedEnd(int colIdx) { return _sortedData[colIdx].end(); }

protected:

   /**
   * A column of the data.
   * The pair represents the index of the example and the value of the column.
   * The index of the column is the index of the vector itself.
   * @remark I am storing both the index and the value because it is a trade off between.
   * speed in a key part of the code (finding the threshold) and the memory consumption.
   * In case of very large databases, this could be turned into a index only vector.
   * @date 11/11/2005
   */
   typedef vector< pair<int, double> > column;

   vector<column>    _sortedData; //!< the sorted data.

};

} // end of namespace MultiBoost

#endif // __SORTED_DATA_H