GUIHMM.h

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) 1999-2008 Soeren Sonnenburg
 * Written (W) 1999-2008 Gunnar Raetsch
 * Copyright (C) 1999-2008 Fraunhofer Institute FIRST and Max-Planck-Society
 */

#ifndef __GUIHMM__H
#define __GUIHMM__H

#include <shogun/lib/config.h>
#include <shogun/base/SGObject.h>
#include <shogun/distributions/HMM.h>
#include <shogun/labels/Labels.h>
#include <shogun/labels/RegressionLabels.h>

namespace shogun
{
class CSGInterface;

class CGUIHMM : public CSGObject
{
    public:
        CGUIHMM() { };
        CGUIHMM(CSGInterface* interface);
        ~CGUIHMM();

        bool new_hmm(int32_t n, int32_t m);
        bool load(char* filename);
        bool save(char* filename, bool is_binary=false);

        bool set_num_hmm_tables(char* param) ;
        bool baum_welch_train();
        bool baum_welch_trans_train();
        bool baum_welch_train_defined();
        bool viterbi_train_defined();
        bool viterbi_train();
        bool linear_train(char align='l');
        bool linear_train_from_file(char* param);
        bool append_model(char* filename, int32_t base1=-1, int32_t base2=-1);
        bool add_states(int32_t num_states=1, float64_t value=0);
        bool set_hmm_as(char* target);
        bool set_pseudo(float64_t pseudo);
        bool convergence_criteria(
            int32_t num_iterations=100, float64_t epsilon=0.001);
        bool output_hmm();
        bool output_hmm_defined();
        bool best_path(int32_t from=0, int32_t to=100);
        bool normalize(bool keep_dead_states=false);
        bool save_path(char* filename, bool is_binary=false);
        bool save_likelihood(char* filename, bool is_binary=false);
        bool load_definitions(char* filename, bool do_init=false);
        bool set_max_dim(char* param);
        bool likelihood();
        bool chop(float64_t value);
        bool relative_entropy(float64_t*& values, int32_t& len);
        bool entropy(float64_t*& values, int32_t& len);
        bool permutation_entropy(int32_t width=0, int32_t seq_num=-1);
        inline CHMM* get_pos() { return pos; }
        inline CHMM* get_neg() { return neg; }
        inline CHMM* get_test() { return test; }
        inline void set_current(CHMM* h) { working=h; }
        inline CHMM* get_current() { return working; }
        inline float64_t get_pseudo() { return PSEUDO; }

        CRegressionLabels* classify(CRegressionLabels* output=NULL);
        float64_t classify_example(int32_t idx);
        CRegressionLabels* one_class_classify(CRegressionLabels* output=NULL);
        CRegressionLabels* linear_one_class_classify(CRegressionLabels* output=NULL);
        float64_t one_class_classify_example(int32_t idx);

        virtual const char* get_name() const { return "GUIHMM"; }

    protected:
        bool converge(float64_t x, float64_t y);
        void switch_model(CHMM** m1, CHMM** m2);

        CHMM* working;

        CHMM* pos;
        CHMM* neg;
        CHMM* test;

        float64_t PSEUDO;
        int32_t M;

    protected:
        CSGInterface* ui;
};
}
#endif