/home/rex/ebltrunk/core/libeblearn/include/ebl_trainer.h

/***************************************************************************
 *   Copyright (C) 2008 by Yann LeCun and Pierre Sermanet *
 *   yann@cs.nyu.edu, pierre.sermanet@gmail.com *
 *   All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are met:
 *     * Redistributions of source code must retain the above copyright
 *       notice, this list of conditions and the following disclaimer.
 *     * Redistributions in binary form must reproduce the above copyright
 *       notice, this list of conditions and the following disclaimer in the
 *       documentation and/or other materials provided with the distribution.
 *     * Redistribution under a license not approved by the Open Source
 *       Initiative (http://www.opensource.org) must display the
 *       following acknowledgement in all advertising material:
 *        This product includes software developed at the Courant
 *        Institute of Mathematical Sciences (http://cims.nyu.edu).
 *     * The names of the authors may not be used to endorse or promote products
 *       derived from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 * DISCLAIMED. IN NO EVENT SHALL ThE AUTHORS BE LIABLE FOR ANY
 * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 ***************************************************************************/

#ifndef EBL_TRAINER_H_
#define EBL_TRAINER_H_

#include "libidx.h"
#include "ebl_arch.h"
#include "ebl_machines.h"
#include "ebl_logger.h"
#include "datasource.h"
#include "ebl_answer.h"

namespace ebl {

  template<typename Tnet, typename Tdata, typename Tlabel>
    class supervised_trainer {
  public:
    supervised_trainer(trainable_module<Tnet,Tdata,Tlabel> &m,
                       parameter<Tnet,bbstate_idx<Tnet> > &p);
    virtual ~supervised_trainer();

    // per-sample methods //////////////////////////////////////////////////////

    bool test_sample(labeled_datasource<Tnet,Tdata,Tlabel> &ds,
                     bbstate_idx<Tnet> &label, bbstate_idx<Tnet> &answers,
                     infer_param &infp);
    Tnet train_sample(labeled_datasource<Tnet,Tdata,Tlabel> &ds,
                      gd_param &arg);

    // epoch methods ///////////////////////////////////////////////////////////

    void test(labeled_datasource<Tnet, Tdata, Tlabel> &ds,
              classifier_meter &log, infer_param &infp,
              uint max_test = 0);
    void train(labeled_datasource<Tnet, Tdata, Tlabel> &ds,
               classifier_meter &log, gd_param &args, int niter,
               infer_param &infp, 
               intg hessian_period = 0, intg nhessian = 0, double mu = .02);
    void compute_diaghessian(labeled_datasource<Tnet, Tdata, Tlabel> &ds,
                             intg niter, double mu);

    // accessors ///////////////////////////////////////////////////////////////
    
    void set_iteration(int i);
    void pretty(labeled_datasource<Tnet, Tdata, Tlabel> &ds);
    void set_progress_file(const std::string &s);
    void update_progress();

    // friends /////////////////////////////////////////////////////////////////
    
    // template <class Tdata, class Tlabel> friend class supervised_trainer_gui;
    template <class T1, class T2, class T3> friend class supervised_trainer_gui;

    // internal methods ////////////////////////////////////////////////////////
  protected:
    
    void init(labeled_datasource<Tnet, Tdata, Tlabel> &ds,
              classifier_meter *log = NULL, bool new_iteration = false);

    // members /////////////////////////////////////////////////////////////////
  protected:
    trainable_module<Tnet,Tdata,Tlabel> &machine;
    parameter<Tnet, bbstate_idx<Tnet> >   &param;       
    bbstate_idx<Tnet>    energy;        
    bbstate_idx<Tnet>   *answers;       
    bbstate_idx<Tnet>   *label;         
    intg                 age;
    int                  iteration;
    void                *iteration_ptr;
    bool                 prettied;      
    std::string          progress_file; 
    intg                 progress_cnt;  
    bool                 test_running;
  };

} // namespace ebl {

#include "ebl_trainer.hpp"

#endif /* EBL_TRAINER_H_ */