PLearn 0.1
NeighborhoodBoxVolumeDensityEstimator.cc
Go to the documentation of this file.
00001 // -*- C++ -*-
00002 
00003 // NeighborhoodBoxVolumeDensityEstimator.cc
00004 //
00005 // Copyright (C) 2006 Pascal Vincent
00006 //
00007 // Redistribution and use in source and binary forms, with or without
00008 // modification, are permitted provided that the following conditions are met:
00009 //
00010 //  1. Redistributions of source code must retain the above copyright
00011 //     notice, this list of conditions and the following disclaimer.
00012 //
00013 //  2. Redistributions in binary form must reproduce the above copyright
00014 //     notice, this list of conditions and the following disclaimer in the
00015 //     documentation and/or other materials provided with the distribution.
00016 //
00017 //  3. The name of the authors may not be used to endorse or promote
00018 //     products derived from this software without specific prior written
00019 //     permission.
00020 //
00021 // THIS SOFTWARE IS PROVIDED BY THE AUTHORS ``AS IS'' AND ANY EXPRESS OR
00022 // IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
00023 // OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN
00024 // NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
00025 // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
00026 // TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
00027 // PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
00028 // LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
00029 // NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
00030 // SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
00031 //
00032 // This file is part of the PLearn library. For more information on the PLearn
00033 // library, go to the PLearn Web site at www.plearn.org
00034 
00035 /* *******************************************************
00036  * $Id: NeighborhoodBoxVolumeDensityEstimator.cc 5668 2006-05-25 17:50:24Z plearner $
00037  ******************************************************* */
00038 
00039 // Authors: Pascal Vincent
00040 
00044 #include "NeighborhoodBoxVolumeDensityEstimator.h"
00045 #include <plearn/vmat/ConcatColumnsVMatrix.h>
00046 #include <plearn/vmat/MemoryVMatrix.h>
00047 #include <plearn_learners/nearest_neighbors/ExhaustiveNearestNeighbors.h>
00048 #include <plearn/base/tostring.h>
00049 
00050 namespace PLearn {
00051 using namespace std;
00052 
00054 // NeighborhoodBoxVolumeDensityEstimator //
00056 NeighborhoodBoxVolumeDensityEstimator::NeighborhoodBoxVolumeDensityEstimator()
00057     :nneighbors(1),
00058      min_radius(1e-6)
00059 {
00060     // for default use Exhaustive search and default Euclidean distance
00061     NN = new ExhaustiveNearestNeighbors(); 
00062 }
00063 
00064 PLEARN_IMPLEMENT_OBJECT(NeighborhoodBoxVolumeDensityEstimator,
00065                         "Simple density estimation based on the volume of the smallest symmetic box, "
00066                         "centered on the test point, and containing its k neighbors.",
00067                         "This estimator is not guaranteed to sum to 1."
00068     );
00069 
00071 // declareOptions //
00073 void NeighborhoodBoxVolumeDensityEstimator::declareOptions(OptionList& ol)
00074 {
00075     // ### Declare all of this object's options here
00076     // ### For the "flags" of each option, you should typically specify
00077     // ### one of OptionBase::buildoption, OptionBase::learntoption or
00078     // ### OptionBase::tuningoption. Another possible flag to be combined with
00079     // ### is OptionBase::nosave
00080 
00081     declareOption(ol, "nneighbors", &NeighborhoodBoxVolumeDensityEstimator::nneighbors, OptionBase::buildoption,
00082                   "Number of neighbors to consider to determine the boxed region centered on the test point");
00083 
00084     declareOption(ol, "min_radius", &NeighborhoodBoxVolumeDensityEstimator::min_radius, OptionBase::buildoption,
00085                   "This is added to each dimension of the found box (as some dimensions may be initially 0)");
00086 
00087     declareOption(ol, "train_set", &NeighborhoodBoxVolumeDensityEstimator::train_set, OptionBase::learntoption,
00088                   "We need to store the training set, as this learner is memory-based...");
00089 
00090     declareOption(ol, "NN", &NeighborhoodBoxVolumeDensityEstimator::NN, OptionBase::buildoption,
00091                   "The nearest neighbor search method to use (default uses ehaustive search with Euclidean distance)");
00092 
00093     // Now call the parent class' declareOptions().
00094     inherited::declareOptions(ol);
00095 }
00096 
00098 // build //
00100 void NeighborhoodBoxVolumeDensityEstimator::build()
00101 {
00102     // ### Nothing to add here, simply calls build_().
00103     inherited::build();
00104     build_();
00105 }
00106 
00108 // build_ //
00110 void NeighborhoodBoxVolumeDensityEstimator::build_()
00111 {
00112     // ### This method should do the real building of the object,
00113     // ### according to set 'options', in *any* situation.
00114     // ### Typical situations include:
00115     // ###  - Initial building of an object from a few user-specified options
00116     // ###  - Building of a "reloaded" object: i.e. from the complete set of all serialised options.
00117     // ###  - Updating or "re-building" of an object after a few "tuning" options have been modified.
00118     // ### You should assume that the parent class' build_() has already been called.
00119 
00120     // ### If the distribution is conditional, you should finish build_() by:
00121     // PDistribution::finishConditionalBuild();
00122 
00123     NN->num_neighbors = nneighbors;
00124     NN->copy_input = false;
00125     NN->copy_target = false;
00126     NN->copy_weight = false;
00127     NN->copy_index = true;
00128     NN->build();
00129     if(!train_set.isNull())
00130     {
00131         NN->setTrainingSet(train_set);
00132         NN->train();
00133     }
00134     NN_outputs.resize(nneighbors);
00135 }
00136 
00138 // log_density //
00140 bool NeighborhoodBoxVolumeDensityEstimator::box_contains(const Vec& center, const Vec& radius, const Vec& input) const
00141 {
00142     int w = center.length();
00143     for(int j=0; j<w; j++)
00144         if(fabs(input[j]-center[j])>radius[j])
00145             return false;
00146     return true;
00147 }
00148 
00149 
00150 real NeighborhoodBoxVolumeDensityEstimator::log_density(const Vec& y) const
00151 {
00152     int l = train_set.length();
00153     int w = inputsize();
00154     int ws = train_set->weightsize();
00155     trainsample.resize(w+ws);
00156     Vec input = trainsample.subVec(0,w);
00157 
00158     // Find distance d to width_neighbor neighbour
00159     NN->computeOutput(y, NN_outputs);
00160 
00161     Vec radius(w);
00162     for(int k=0; k<NN_outputs.length(); k++)
00163     {
00164         train_set->getRow((int)NN_outputs[k],trainsample);
00165         for(int j=0; j<w; j++)
00166         {
00167             real d = fabs(y[j]-input[j]);
00168             radius[j] = max(radius[j],d);
00169         }
00170     }
00171     radius += min_radius;
00172     
00173     int region_count = 0;
00174     for(int i=0; i<l; i++)
00175     {
00176         train_set->getRow(i,trainsample);
00177         if(box_contains(y,radius,input))
00178             region_count++;
00179     }
00180     
00181     double log_region_volume = 0;
00182     for(int j=0; j<w; j++)
00183         log_region_volume += pl_log(radius[j]);
00184 
00185     return pl_log(double(region_count))-log_region_volume-pl_log(double(l));
00186 }
00187 
00189 // makeDeepCopyFromShallowCopy //
00191 void NeighborhoodBoxVolumeDensityEstimator::makeDeepCopyFromShallowCopy(CopiesMap& copies)
00192 {
00193     inherited::makeDeepCopyFromShallowCopy(copies);
00194 
00195     // ### Call deepCopyField on all "pointer-like" fields
00196     // ### that you wish to be deepCopied rather than
00197     // ### shallow-copied.
00198     // ### ex:
00199     // deepCopyField(trainvec, copies);
00200     deepCopyField(NN, copies);
00201 }
00202 
00203 
00204 // ### Remove this method, if your distribution does not implement it.
00206 // train //
00208 void NeighborhoodBoxVolumeDensityEstimator::train()
00209 {
00210     NN->setTrainingSet(train_set);
00211     NN->train();
00212 }
00213 
00214 
00215 void NeighborhoodBoxVolumeDensityEstimator::forget()
00216 {
00217     NN->forget();
00218 }
00219 
00220 
00221 }
00222 
00223 
00224 /*
00225   Local Variables:
00226   mode:c++
00227   c-basic-offset:4
00228   c-file-style:"stroustrup"
00229   c-file-offsets:((innamespace . 0)(inline-open . 0))
00230   indent-tabs-mode:nil
00231   fill-column:79
00232   End:
00233 */
00234 // 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