paramEstimator.h

Go to the documentation of this file.

#ifndef GUM_LEARNING_PARAM_ESTIMATOR_H
#define GUM_LEARNING_PARAM_ESTIMATOR_H

#include <type_traits>

#include <agrum/agrum.h>
#include <agrum/learning/database/databaseTable.h>
#include <agrum/learning/aprioris/apriori.h>
#include <agrum/learning/scores_and_tests/recordCounter.h>
#include <agrum/multidim/potential.h>

namespace gum {

  namespace learning {


    template < template < typename > class ALLOC = std::allocator >
    class ParamEstimator {
      public:
      using allocator_type = ALLOC< NodeId >;

      // ##########################################################################
      // ##########################################################################


      ParamEstimator(
         const DBRowGeneratorParser< ALLOC >& parser,
         const Apriori< ALLOC >&              external_apriori,
         const Apriori< ALLOC >&              score_internal__apriori,
         const std::vector< std::pair< std::size_t, std::size_t >,
                            ALLOC< std::pair< std::size_t, std::size_t > > >&
            ranges,
         const Bijection< NodeId, std::size_t, ALLOC< std::size_t > >&
            nodeId2columns =
               Bijection< NodeId, std::size_t, ALLOC< std::size_t > >(),
         const allocator_type& alloc = allocator_type());


      ParamEstimator(
         const DBRowGeneratorParser< ALLOC >& parser,
         const Apriori< ALLOC >&              external_apriori,
         const Apriori< ALLOC >&              score_internal__apriori,
         const Bijection< NodeId, std::size_t, ALLOC< std::size_t > >&
            nodeId2columns =
               Bijection< NodeId, std::size_t, ALLOC< std::size_t > >(),
         const allocator_type& alloc = allocator_type());

      ParamEstimator(const ParamEstimator< ALLOC >& from);

      ParamEstimator(const ParamEstimator< ALLOC >& from,
                     const allocator_type&          alloc);

      ParamEstimator(ParamEstimator< ALLOC >&& from);

      ParamEstimator(ParamEstimator< ALLOC >&& from, const allocator_type& alloc);

      virtual ParamEstimator< ALLOC >* clone() const = 0;

      virtual ParamEstimator< ALLOC >*
         clone(const allocator_type& alloc) const = 0;

      virtual ~ParamEstimator();



      // ##########################################################################
      // ##########################################################################

      virtual void clear();

      virtual void setMaxNbThreads(std::size_t nb) const;

      virtual std::size_t nbThreads() const;

      virtual void setMinNbRowsPerThread(const std::size_t nb) const;

      virtual std::size_t minNbRowsPerThread() const;


      template < template < typename > class XALLOC >
      void setRanges(
         const std::vector< std::pair< std::size_t, std::size_t >,
                            XALLOC< std::pair< std::size_t, std::size_t > > >&
            new_ranges);

      void clearRanges();

      const std::vector< std::pair< std::size_t, std::size_t >,
                         ALLOC< std::pair< std::size_t, std::size_t > > >&
         ranges() const;

      std::vector< double, ALLOC< double > > parameters(const NodeId target_node);


      virtual std::vector< double, ALLOC< double > > parameters(
         const NodeId                                  target_node,
         const std::vector< NodeId, ALLOC< NodeId > >& conditioning_nodes) = 0;


      template < typename GUM_SCALAR >
      void setParameters(
         const NodeId                                  target_node,
         const std::vector< NodeId, ALLOC< NodeId > >& conditioning_nodes,
         Potential< GUM_SCALAR >&                      pot);


      const Bijection< NodeId, std::size_t, ALLOC< std::size_t > >&
         nodeId2Columns() const;

      const DatabaseTable< ALLOC >& database() const;


      template < typename GUM_SCALAR >
      void setBayesNet(const BayesNet< GUM_SCALAR >& new_bn);

      allocator_type getAllocator() const;


      protected:
      Apriori< ALLOC >* _external_apriori{nullptr};

      Apriori< ALLOC >* _score_internal_apriori{nullptr};

      RecordCounter< ALLOC > _counter;

      const std::vector< NodeId, ALLOC< NodeId > > _empty_nodevect;


      ParamEstimator< ALLOC >& operator=(const ParamEstimator< ALLOC >& from);

      ParamEstimator< ALLOC >& operator=(ParamEstimator< ALLOC >&& from);

      private:
#ifndef DOXYGEN_SHOULD_SKIP_THIS

      template < typename GUM_SCALAR >
      void __checkParameters(
         const NodeId                                  target_node,
         const std::vector< NodeId, ALLOC< NodeId > >& conditioning_nodes,
         Potential< GUM_SCALAR >&                      pot);

      // sets the CPT's parameters corresponding to a given Potential
      // when the potential belongs to a BayesNet<GUM_SCALAR> when
      // GUM_SCALAR is different from a double
      template < typename GUM_SCALAR >
      typename std::enable_if< !std::is_same< GUM_SCALAR, double >::value,
                               void >::type
         __setParameters(
            const NodeId                                  target_node,
            const std::vector< NodeId, ALLOC< NodeId > >& conditioning_nodes,
            Potential< GUM_SCALAR >&                      pot);

      // sets the CPT's parameters corresponding to a given Potential
      // when the potential belongs to a BayesNet<GUM_SCALAR> when
      // GUM_SCALAR is equal to double (the code is optimized for doubles)
      template < typename GUM_SCALAR >
      typename std::enable_if< std::is_same< GUM_SCALAR, double >::value,
                               void >::type
         __setParameters(
            const NodeId                                  target_node,
            const std::vector< NodeId, ALLOC< NodeId > >& conditioning_nodes,
            Potential< GUM_SCALAR >&                      pot);

#endif /* DOXYGEN_SHOULD_SKIP_THIS */
    };

  } /* namespace learning */

} /* namespace gum */

#include <agrum/learning/paramUtils/paramEstimator_tpl.h>

#endif /* GUM_LEARNING_PARAM_ESTIMATOR_H */