AveragedPerceptron.cpp

Go to the documentation of this file.

/*
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 3 of the License, or
 * (at your option) any later version.
 *
 * Written (W) 2011 Hidekazu Oiwa
 */

#include <shogun/classifier/AveragedPerceptron.h>
#include <shogun/labels/Labels.h>
#include <shogun/mathematics/Math.h>
#include <shogun/labels/BinaryLabels.h>
#include <shogun/lib/Signal.h>

using namespace shogun;

CAveragedPerceptron::CAveragedPerceptron()
: CLinearMachine(), learn_rate(0.1), max_iter(1000)
{
}

CAveragedPerceptron::CAveragedPerceptron(CDotFeatures* traindat, CLabels* trainlab)
: CLinearMachine(), learn_rate(.1), max_iter(1000)
{
    set_features(traindat);
    set_labels(trainlab);
}

CAveragedPerceptron::~CAveragedPerceptron()
{
}

bool CAveragedPerceptron::train_machine(CFeatures* data)
{
    ASSERT(m_labels)
    ASSERT(m_labels->get_label_type() == LT_BINARY)

    if (data)
    {
        if (!data->has_property(FP_DOT))
            SG_ERROR("Specified features are not of type CDotFeatures\n")
        set_features((CDotFeatures*) data);
    }
    ASSERT(features)
    bool converged=false;
    int32_t iter=0;
    SGVector<int32_t> train_labels=((CBinaryLabels*) m_labels)->get_int_labels();
    int32_t num_feat=features->get_dim_feature_space();
    int32_t num_vec=features->get_num_vectors();

    ASSERT(num_vec==train_labels.vlen)
    SGVector<float64_t> w(num_feat);
    float64_t* tmp_w=SG_MALLOC(float64_t, num_feat);
    memset(tmp_w, 0, sizeof(float64_t)*num_feat);
    float64_t* output=SG_MALLOC(float64_t, num_vec);

    //start with uniform w, bias=0, tmp_bias=0
    bias=0;
    float64_t tmp_bias=0;
    for (int32_t i=0; i<num_feat; i++)
        w[i]=1.0/num_feat;

    //loop till we either get everything classified right or reach max_iter

    while (!(cancel_computation()) && (!converged && iter < max_iter))
    {
        converged=true;
        SG_INFO("Iteration Number : %d of max %d\n", iter, max_iter);

        for (int32_t i=0; i<num_vec; i++)
        {
            output[i] = features->dense_dot(i, w.vector, w.vlen) + bias;

            if (CMath::sign<float64_t>(output[i]) != train_labels.vector[i])
            {
                converged=false;
                bias+=learn_rate*train_labels.vector[i];
                features->add_to_dense_vec(learn_rate*train_labels.vector[i], i, w.vector, w.vlen);
            }

            // Add current w to tmp_w, and current bias to tmp_bias
            // To calculate the sum of each iteration's w, bias
            for (int32_t j=0; j<num_feat; j++)
                tmp_w[j]+=w[j];
            tmp_bias+=bias;
        }
        iter++;
    }

    if (converged)
        SG_INFO("Averaged Perceptron algorithm converged after %d iterations.\n", iter)
    else
        SG_WARNING("Averaged Perceptron algorithm did not converge after %d iterations.\n", max_iter)

    // calculate and set the average paramter of w, bias
    for (int32_t i=0; i<num_feat; i++)
        w[i]=tmp_w[i]/(num_vec*iter);
    bias=tmp_bias/(num_vec*iter);

    SG_FREE(output);
    SG_FREE(tmp_w);

    set_w(w);

    return converged;
}