PLearn 0.1
RegressionTreeMulticlassLeaveFast.cc
Go to the documentation of this file.
00001 // -*- C++ -*-
00002 
00003 // RegressionTreeMulticlassLeaveFast.cc
00004 // Copyright (c) 1998-2002 Pascal Vincent
00005 // Copyright (C) 1999-2002 Yoshua Bengio and University of Montreal
00006 // Copyright (c) 2002 Jean-Sebastien Senecal, Xavier Saint-Mleux, Rejean Ducharme
00007 //
00008 // Redistribution and use in source and binary forms, with or without
00009 // modification, are permitted provided that the following conditions are met:
00010 // 
00011 //  1. Redistributions of source code must retain the above copyright
00012 //     notice, this list of conditions and the following disclaimer.
00013 // 
00014 //  2. Redistributions in binary form must reproduce the above copyright
00015 //     notice, this list of conditions and the following disclaimer in the
00016 //     documentation and/or other materials provided with the distribution.
00017 // 
00018 //  3. The name of the authors may not be used to endorse or promote
00019 //     products derived from this software without specific prior written
00020 //     permission.
00021 // 
00022 // THIS SOFTWARE IS PROVIDED BY THE AUTHORS ``AS IS'' AND ANY EXPRESS OR
00023 // IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
00024 // OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN
00025 // NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
00026 // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
00027 // TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
00028 // PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
00029 // LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
00030 // NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
00031 // SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
00032 // 
00033 // This file is part of the PLearn library. For more information on the PLearn
00034 // library, go to the PLearn Web site at www.plearn.org
00035 
00036 
00037 /* ********************************************************************************    
00038  * $Id: RegressionTreeMulticlassLeaveFast.cc, v 1.0 2004/07/19 10:00:00 Bengio/Kegl/Godbout    *
00039  * This file is part of the PLearn library.                                     *
00040  ******************************************************************************** */
00041 
00042 #include "RegressionTreeMulticlassLeaveFast.h"
00043 #include "RegressionTreeRegisters.h"
00044 #include <plearn/math/TMat_maths_impl.h>
00045 
00046 namespace PLearn {
00047 using namespace std;
00048 
00049 PLEARN_IMPLEMENT_OBJECT(RegressionTreeMulticlassLeaveFast,
00050                         "Object to represent the leaves of a regression tree.",
00051                         "It maintains the necessary statistics to compute the output and the train error\n"
00052                         "of the samples in the leave.\n"
00053     );
00054 
00055 RegressionTreeMulticlassLeaveFast::RegressionTreeMulticlassLeaveFast()
00056     : nb_class(-1),
00057       objective_function("l1")
00058 {
00059     build();
00060 }
00061 
00062 RegressionTreeMulticlassLeaveFast::~RegressionTreeMulticlassLeaveFast()
00063 {
00064 }
00065 
00066 void RegressionTreeMulticlassLeaveFast::declareOptions(OptionList& ol)
00067 { 
00068     inherited::declareOptions(ol);
00069 
00070     declareOption(ol, "nb_class", 
00071                   &RegressionTreeMulticlassLeaveFast::nb_class,
00072                   OptionBase::buildoption,
00073                   "The number of class. Should be numbered from 0 to nb_class -1.\n"
00074                   );
00075     declareOption(ol, "objective_function",
00076                   &RegressionTreeMulticlassLeaveFast::objective_function,
00077                   OptionBase::buildoption,
00078                   "The function to be used to compute the leave error.\n"
00079                   "Current supported values are l1 and l2 (default is l1).");
00080       
00081     declareOption(ol, "multiclass_weights_sum",
00082                   &RegressionTreeMulticlassLeaveFast::multiclass_weights_sum,
00083                   OptionBase::learntoption,
00084                   "A vector to count the weight sum of each possible output "
00085                   "for the sample in this leave.\n");
00086     redeclareOption(ol, "loss_function_factor",
00087                   &RegressionTreeMulticlassLeaveFast::loss_function_factor,
00088                   OptionBase::learntoption,
00089                   "The loss fct factor. Depend of the objective_function.\n");
00090 }
00091 
00092 void RegressionTreeMulticlassLeaveFast::makeDeepCopyFromShallowCopy(CopiesMap& copies)
00093 {
00094     inherited::makeDeepCopyFromShallowCopy(copies);
00095     deepCopyField(objective_function, copies);
00096     deepCopyField(multiclass_weights_sum, copies);
00097 }
00098 
00099 void RegressionTreeMulticlassLeaveFast::build()
00100 {
00101     inherited::build();
00102     build_();
00103 }
00104 
00105 void RegressionTreeMulticlassLeaveFast::build_()
00106 {
00107 }
00108 
00109 void RegressionTreeMulticlassLeaveFast::initStats()
00110 {
00111     length_ = 0;
00112     weights_sum = 0.0;
00113     if (loss_function_weight != 0.0)
00114     {
00115         if(objective_function == "l1")
00116             loss_function_factor = 2.0 / loss_function_weight;
00117         else
00118             loss_function_factor = 2.0 / pow(loss_function_weight, 2);
00119     }
00120     else
00121     {
00122         loss_function_factor = 1.0;
00123     }
00124     multiclass_weights_sum.resize(nb_class);
00125     multiclass_weights_sum.fill(0);
00126 }
00127 
00128 void RegressionTreeMulticlassLeaveFast::addRow(int row)
00129 {
00130     real weight = train_set->getWeight(row);
00131     real target = train_set->getTarget(row);
00132     RegressionTreeMulticlassLeaveFast::addRow(row, target, weight);
00133 }
00134 
00135 void RegressionTreeMulticlassLeaveFast::addRow(int row, real target, real weight,
00136                                  Vec outputv, Vec errorv)
00137 {
00138     RegressionTreeMulticlassLeaveFast::addRow(row, target, weight);
00139     RegressionTreeMulticlassLeaveFast::getOutputAndError(outputv,errorv);
00140 }
00141 
00142 void RegressionTreeMulticlassLeaveFast::addRow(int row, real target, real weight)
00143 {
00144     length_ += 1;
00145     weights_sum += weight;
00146     multiclass_weights_sum[int(target)] += weight;
00147 }
00148 
00149 void RegressionTreeMulticlassLeaveFast::addRow(int row, Vec outputv, Vec errorv)
00150 {
00151     RegressionTreeMulticlassLeaveFast::addRow(row);
00152     RegressionTreeMulticlassLeaveFast::getOutputAndError(outputv,errorv);    
00153 }
00154 
00155 void RegressionTreeMulticlassLeaveFast::removeRow(int row, Vec outputv, Vec errorv)
00156 {
00157     real weight = train_set->getWeight(row);
00158     real target = train_set->getTarget(row);
00159     RegressionTreeMulticlassLeaveFast::removeRow(row,target,weight,outputv,errorv);
00160 }
00161 
00162 void RegressionTreeMulticlassLeaveFast::removeRow(int row, real target, real weight,
00163                                  Vec outputv, Vec errorv){
00164     RegressionTreeMulticlassLeaveFast::removeRow(row,target,weight);
00165     RegressionTreeMulticlassLeaveFast::getOutputAndError(outputv,errorv);
00166 }
00167 
00168 void RegressionTreeMulticlassLeaveFast::removeRow(int row, real target, real weight)
00169 {
00170     length_ -= 1;
00171     weights_sum -= weight;
00172     PLASSERT(length_>=0);
00173     PLASSERT(weights_sum>=0);
00174     PLASSERT(length_>0 || weights_sum==0);
00175     multiclass_weights_sum[int(target)] -= weight;
00176 }
00177 
00178 void RegressionTreeMulticlassLeaveFast::getOutputAndError(Vec& output, Vec& error)const
00179 {
00180 #ifdef BOUNDCHECK
00181     if(nb_class<=0)
00182         PLERROR("In RegressionTreeMulticlassLeaveFast::getOutputAndError() -"
00183                 " nb_class must be set.");
00184 #endif
00185     if(length_==0){        
00186         output.clear();
00187         output[0]=MISSING_VALUE;
00188         error.clear();
00189         return;
00190     }
00191     int mc_winer = 0;
00192     real conf = 0;
00193     //index of the max. Is their an optimized version?
00194     for (int mc_ind = 1; mc_ind < nb_class; mc_ind++)
00195     {
00196         if (multiclass_weights_sum[mc_ind] > multiclass_weights_sum[mc_winer])
00197             mc_winer = mc_ind;
00198     }
00199     output[0] = mc_winer;
00200     if (missing_leave)
00201     {
00202         error[0] = 0.0;
00203         error[1] = weights_sum;
00204         error[2] = 0.0;
00205     }
00206     else
00207     {
00208         conf = multiclass_weights_sum[mc_winer] / weights_sum;
00209         error[0] = 0.0;
00210         if (objective_function == "l1")
00211         {
00212             for (int mc_ind = 0; mc_ind < nb_class;mc_ind++)
00213             {
00214                 error[0] += abs(mc_winer - mc_ind) 
00215                     * multiclass_weights_sum[mc_ind];
00216             }
00217         }
00218         else
00219         {
00220             for (int mc_ind = 0; mc_ind < nb_class;mc_ind++)
00221             {
00222                 error[0] += pow(mc_winer - mc_ind, 2.) 
00223                     * multiclass_weights_sum[mc_ind];
00224             }
00225         }
00226         error[0] *= loss_function_factor * length_ / weights_sum;
00227         if (error[0] < 1E-10) error[0] = 0.0;
00228         if (error[0] > weights_sum * loss_function_factor)
00229             error[2] = weights_sum * loss_function_factor;
00230         else error[2] = error[0];
00231         error[1] = (1.0 - conf) * length_;
00232     }
00233     if(output_confidence_target) output[1] = conf;
00234 }
00235 
00236 TVec<string> RegressionTreeMulticlassLeaveFast::getOutputNames() const
00237 {
00238     TVec<string> ret;
00239     ret.append("class_pred");
00240     if(output_confidence_target)
00241         ret.append("confidence");
00242     return ret;
00243 }
00244 
00245 void RegressionTreeMulticlassLeaveFast::addLeave(PP<RegressionTreeLeave> leave_){
00246     PP<RegressionTreeMulticlassLeaveFast> leave = (PP<RegressionTreeMulticlassLeaveFast>) leave_;
00247 
00248     if(leave->classname() == classname()){
00249         length_ += leave->length_;
00250         weights_sum += leave->weights_sum;
00251         multiclass_weights_sum += leave->multiclass_weights_sum;
00252     }else
00253         PLERROR("In %s::addLeave the leave to add should have the same class. It have %s.",
00254                 classname().c_str(), leave->classname().c_str());
00255 }
00256 
00257 void RegressionTreeMulticlassLeaveFast::removeLeave(PP<RegressionTreeLeave> leave_){
00258     PP<RegressionTreeMulticlassLeaveFast> leave = (PP<RegressionTreeMulticlassLeaveFast>) leave_;
00259 
00260     if(leave->classname() == classname()){
00261         length_ -= leave->length_;
00262         weights_sum -= leave->weights_sum;
00263         multiclass_weights_sum -= leave->multiclass_weights_sum;
00264     }else
00265         PLERROR("In %s::addLeave the leave to add should have the same class. It have %s.",
00266                 classname().c_str(), leave->classname().c_str());
00267 }
00268 
00269 void RegressionTreeMulticlassLeaveFast::printStats()
00270 {
00271     cout << " l " << length_;
00272     Vec output(2);
00273     Vec error(3);
00274     getOutputAndError(output,error);
00275     cout << " o0 " << output[0];
00276     cout << " o1 " << output[1];
00277     cout << " e0 " << error[0];
00278     cout << " e1 " << error[1];
00279     cout << " ws " << weights_sum;
00280     cout << endl;
00281     cout << " mws " << multiclass_weights_sum << endl;
00282 }
00283 
00284 } // end of namespace PLearn
00285 
00286 
00287 /*
00288   Local Variables:
00289   mode:c++
00290   c-basic-offset:4
00291   c-file-style:"stroustrup"
00292   c-file-offsets:((innamespace . 0)(inline-open . 0))
00293   indent-tabs-mode:nil
00294   fill-column:79
00295   End:
00296 */
00297 // vim: filetype=cpp:expandtab:shiftwidth=4:tabstop=8:softtabstop=4:encoding=utf-8:textwidth=79 :
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Defines