/usr/include/shogun/multiclass/MulticlassOneVsOneStrategy.h is in libshogun-dev 3.2.0-7.3build4.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 | /*
* 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) 2012 Chiyuan Zhang
* Written (W) 2013 Shell Hu and Heiko Strathmann
* Copyright (C) 2012 Chiyuan Zhang
*/
#include <shogun/multiclass/MulticlassStrategy.h>
namespace shogun
{
/** @brief multiclass one vs one strategy
* used to train generic multiclass machines
* for K-class problems with building
* voting-based ensemble of K*(K-1)
* binary classifiers
* multiclass probabilistic outputs can be
* obtained by using the heuristics described in [1]
*
* [1] J. Milgram, M. Cheriet, R.Sabourin, "One Against One" or "One Against One":
* Which One is Better for Handwriting Recognition with SVMs?
*/
class CMulticlassOneVsOneStrategy: public CMulticlassStrategy
{
public:
/** constructor */
CMulticlassOneVsOneStrategy();
/** constructor
* @param prob_heuris probability estimation heuristic
*/
CMulticlassOneVsOneStrategy(EProbHeuristicType prob_heuris);
/** destructor */
virtual ~CMulticlassOneVsOneStrategy() {}
/** start training */
virtual void train_start(CMulticlassLabels *orig_labels, CBinaryLabels *train_labels);
/** has more training phase */
virtual bool train_has_more();
/** prepare for the next training phase.
* @return the subset that should be applied before training.
*/
virtual SGVector<int32_t> train_prepare_next();
/** decide the final label.
* @param outputs a vector of output from each machine (in that order)
*/
virtual int32_t decide_label(SGVector<float64_t> outputs);
/** get number of machines used in this strategy.
*/
virtual int32_t get_num_machines()
{
return m_num_classes*(m_num_classes-1)/2;
}
/** get name */
virtual const char* get_name() const
{
return "MulticlassOneVsOneStrategy";
};
/** rescale multiclass outputs according to the selected heuristic
* @param outputs a vector of output from each machine (in that order)
* which will be resized to length of num_classes if heuristic is set
*/
virtual void rescale_outputs(SGVector<float64_t> outputs);
/** set the number of classes, since the number of machines totally
* depends on the number of classes, which will also be set.
* @param num_classes number of classes
*/
void set_num_classes(int32_t num_classes)
{
CMulticlassStrategy::set_num_classes(num_classes);
m_num_machines = m_num_classes*(m_num_classes-1)/2;
}
protected:
/** OVO Price's heuristic see [1]
* @param outputs a vector of output from each machine (in that order)
* @param indx1 indices of 1st involved class of training machines
* @param indx2 indices of 2nd involved class of training machines
*/
void rescale_heuris_price(SGVector<float64_t> outputs,
const SGVector<int32_t> indx1, const SGVector<int32_t> indx2);
/** OVO Hastie's heuristic see [1]
* @param outputs a vector of output from each machine (in that order)
* @param indx1 indices of 1st involved class of training machines
* @param indx2 indices of 2nd involved class of training machines
*/
void rescale_heuris_hastie(SGVector<float64_t> outputs,
const SGVector<int32_t> indx1, const SGVector<int32_t> indx2);
/** OVO Hamamura's heuristic see [1]
* @param outputs a vector of output from each machine (in that order)
* @param indx1 indices of 1st involved class of training machines
* @param indx2 indices of 2nd involved class of training machines
*/
void rescale_heuris_hamamura(SGVector<float64_t> outputs,
const SGVector<int32_t> indx1, const SGVector<int32_t> indx2);
private:
/** register parameters */
void register_parameters();
protected:
int32_t m_num_machines; ///< number of machines
int32_t m_train_pair_idx_1; ///< 1st index of current submachine being trained
int32_t m_train_pair_idx_2; ///< 2nd index of current submachine being trained
SGVector<int32_t> m_num_samples; ///< number of samples per machine
};
} // namespace shogun
|