scoreBDeu.cpp

Go to the documentation of this file.

/****************************************************************************
 *   This file is part of the aGrUM/pyAgrum library.                        *
 *                                                                          *
 *   Copyright (c) 2005-2026 by                                             *
 *       - Pierre-Henri WUILLEMIN(_at_LIP6)                                 *
 *       - Christophe GONZALES(_at_AMU)                                     *
 *                                                                          *
 *   The aGrUM/pyAgrum library is free software; you can redistribute it    *
 *   and/or modify it under the terms of either :                           *
 *                                                                          *
 *    - the GNU Lesser General Public License as published by               *
 *      the Free Software Foundation, either version 3 of the License,      *
 *      or (at your option) any later version,                              *
 *    - the MIT license (MIT),                                              *
 *    - or both in dual license, as here.                                   *
 *                                                                          *
 *   (see https://agrum.gitlab.io/articles/dual-licenses-lgplv3mit.html)    *
 *                                                                          *
 *   This aGrUM/pyAgrum library is distributed in the hope that it will be  *
 *   useful, but WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,          *
 *   INCLUDING BUT NOT LIMITED TO THE WARRANTIES MERCHANTABILITY or FITNESS *
 *   FOR A PARTICULAR PURPOSE  AND NONINFRINGEMENT. IN NO EVENT SHALL THE   *
 *   AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER *
 *   LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,        *
 *   ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR  *
 *   OTHER DEALINGS IN THE SOFTWARE.                                        *
 *                                                                          *
 *   See LICENCES for more details.                                         *
 *                                                                          *
 *   SPDX-FileCopyrightText: Copyright 2005-2026                            *
 *       - Pierre-Henri WUILLEMIN(_at_LIP6)                                 *
 *       - Christophe GONZALES(_at_AMU)                                     *
 *   SPDX-License-Identifier: LGPL-3.0-or-later OR MIT                      *
 *                                                                          *
 *   Contact  : info_at_agrum_dot_org                                       *
 *   homepage : http://agrum.gitlab.io                                      *
 *   gitlab   : https://gitlab.com/agrumery/agrum                           *
 *                                                                          *
 ****************************************************************************/
 

 
#include <agrum/BN/learning/scores/scoreBDeu.h>
 
#ifndef DOXYGEN_SHOULD_SKIP_THIS

#  ifdef GUM_NO_INLINE
#    include <agrum/BN/learning/scores/scoreBDeu_inl.h>
#  endif /* GUM_NO_INLINE */
 
namespace gum {
 
  namespace learning {
 
    // Constructors and destructor are defined out-of-line (not INLINE) on
    // purpose: see the comment in score.cpp for the MSVC LNK2005 rationale.

    ScoreBDeu::ScoreBDeu(const DBRowGeneratorParser&                                 parser,
                         const Prior&                                                prior,
                         const std::vector< std::pair< std::size_t, std::size_t > >& ranges,
                         const Bijection< NodeId, std::size_t >& nodeId2columns) :
        Score(parser, prior, ranges, nodeId2columns),
        _internal_prior_(parser.database(), nodeId2columns) {
      GUM_CONSTRUCTOR(ScoreBDeu);
    }

    ScoreBDeu::ScoreBDeu(const DBRowGeneratorParser&             parser,
                         const Prior&                            prior,
                         const Bijection< NodeId, std::size_t >& nodeId2columns) :
        Score(parser, prior, nodeId2columns), _internal_prior_(parser.database(), nodeId2columns) {
      GUM_CONSTRUCTOR(ScoreBDeu);
    }

    ScoreBDeu::ScoreBDeu(const ScoreBDeu& from) :
        Score(from), _internal_prior_(from._internal_prior_), _gammalog2_(from._gammalog2_) {
      GUM_CONS_CPY(ScoreBDeu);
    }

    ScoreBDeu::ScoreBDeu(ScoreBDeu&& from) :
        Score(std::move(from)), _internal_prior_(std::move(from._internal_prior_)),
        _gammalog2_(std::move(from._gammalog2_)) {
      GUM_CONS_MOV(ScoreBDeu);
    }

    ScoreBDeu::~ScoreBDeu() { GUM_DESTRUCTOR(ScoreBDeu); }

    ScoreBDeu& ScoreBDeu::operator=(const ScoreBDeu& from) {
      if (this != &from) {
        Score::operator=(from);
        _internal_prior_ = from._internal_prior_;
      }
      return *this;
    }

    ScoreBDeu& ScoreBDeu::operator=(ScoreBDeu&& from) {
      if (this != &from) {
        Score::operator=(std::move(from));
        _internal_prior_ = std::move(from._internal_prior_);
      }
      return *this;
    }

    std::string ScoreBDeu::isPriorCompatible(PriorType prior_type, double weight) {
      // check that the prior is compatible with the score
      if (prior_type == PriorType::NoPriorType) { return ""; }
 
      if (weight == 0.0) {
        return "The prior is currently compatible with the BDeu score but "
               "if you change the weight, it will become incompatible.";
      }
 
      // known incompatible priors
      if ((prior_type == PriorType::DirichletPriorType)
          || (prior_type == PriorType::SmoothingPriorType)) {
        return "The BDeu score already contains a different 'implicit' prior. "
               "Therefore, the learning will probably be biased.";
      }
 
      // prior types unsupported by the type checker
      return std::format("The prior '{}' is not yet compatible with the score 'BDeu'.",
                         priorTypeToString(prior_type));
    }

    double ScoreBDeu::score_(const IdCondSet& idset) {
      // get the counts for all the nodes in the idset and add the prior
      std::vector< double > N_ijk(this->counter_.counts(idset, true));
      const std::size_t     all_size = N_ijk.size();
 
      double       score                      = 0.0;
      const double ess                        = _internal_prior_.weight();
      const bool   informative_external_prior = this->prior_->isInformative();
 
 
      // here, we distinguish idsets with conditioning nodes from those
      // without conditioning nodes
      if (idset.hasConditioningSet()) {
        // get the counts for the conditioning nodes
        std::vector< double > N_ij(this->marginalize_(idset[0], N_ijk));
        const std::size_t     conditioning_size = N_ij.size();
        const double          ess_qi            = ess / conditioning_size;
        const double          ess_riqi          = ess / all_size;
 
        if (informative_external_prior) {
          // the score to compute is that of BD with priors
          // N'_ijk + ESS / (r_i * q_i )
          // (the + ESS / (r_i * q_i ) is here to take into account the
          // internal prior of BDeu)
          std::vector< double > N_prime_ijk(all_size, 0.0);
          this->prior_->addJointPseudoCount(idset, N_prime_ijk);
          std::vector< double > N_prime_ij(N_ij.size(), 0.0);
          this->prior_->addConditioningPseudoCount(idset, N_prime_ij);
 
          // the BDeu score can be computed as follows:
          // sum_j=1^qi [ gammalog2 ( N'_ij + ESS / q_i ) -
          //     gammalog2 ( N_ij + N'_ij + ESS / q_i )
          //     + sum_k=1^ri { gammlog2 ( N_ijk + N'_ijk + ESS / (r_i * q_i ) )
          //     - gammalog2 ( N'_ijk + ESS / (r_i * q_i ) ) } ]
          for (std::size_t j = std::size_t(0); j < conditioning_size; ++j) {
            score += _gammalog2_(N_prime_ij[j] + ess_qi)
                   - _gammalog2_(N_ij[j] + N_prime_ij[j] + ess_qi);
          }
          for (std::size_t k = std::size_t(0); k < all_size; ++k) {
            score += _gammalog2_(N_ijk[k] + N_prime_ijk[k] + ess_riqi)
                   - _gammalog2_(N_prime_ijk[k] + ess_riqi);
          }
        } else {
          // the BDeu score can be computed as follows:
          // qi * gammalog2 (ess / qi) - ri * qi * gammalog2 (ess / (ri * qi) )
          // - sum_j=1^qi [ gammalog2 ( N_ij + ess / qi ) ]
          // + sum_j=1^qi sum_k=1^ri log [ gammalog2 ( N_ijk + ess / (ri * qi) )
          // ]
          score = conditioning_size * _gammalog2_(ess_qi) - all_size * _gammalog2_(ess_riqi);
 
          for (const auto n_ij: N_ij) {
            score -= _gammalog2_(n_ij + ess_qi);
          }
          for (const auto n_ijk: N_ijk) {
            score += _gammalog2_(n_ijk + ess_riqi);
          }
        }
      } else {
        // here, there are no conditioning nodes
        const double ess_ri = ess / all_size;
 
        if (informative_external_prior) {
          // the score to compute is that of BD with priors
          // N'_ijk + ESS / ( ri * qi )
          // (the + ESS / ( ri * qi ) is here to take into account the
          // internal prior of K2)
          std::vector< double > N_prime_ijk(all_size, 0.0);
          this->prior_->addJointPseudoCount(idset, N_prime_ijk);
 
          // the BDeu score can be computed as follows:
          // gammalog2 ( N' + ess ) - gammalog2 ( N + N' + ess )
          // + sum_k=1^ri { gammlog2 ( N_i + N'_i + ESS / ri)
          //                - gammalog2 ( N'_i + ESS / ri ) }
          double N       = 0.0;
          double N_prime = 0.0;
          for (std::size_t k = std::size_t(0); k < all_size; ++k) {
            score += _gammalog2_(N_ijk[k] + N_prime_ijk[k] + ess_ri)
                   - _gammalog2_(N_prime_ijk[k] + ess_ri);
            N += N_ijk[k];
            N_prime += N_prime_ijk[k];
          }
          score += _gammalog2_(N_prime + ess) - _gammalog2_(N + N_prime + ess);
        } else {
          // the BDeu score can be computed as follows:
          // gammalog2 ( ess ) - ri * gammalog2 ( ess / ri )
          // - gammalog2 ( N + ess )
          // + sum_k=1^ri log [ gammalog2 ( N_ijk + ess / ri ) ]
 
          score    = _gammalog2_(ess) - all_size * _gammalog2_(ess_ri);
          double N = 0;
          for (const auto n_ijk: N_ijk) {
            score += _gammalog2_(n_ijk + ess_ri);
            N += n_ijk;
          }
          score -= _gammalog2_(N + ess);
        }
      }
 
      return score;
    }
 
  } /* namespace learning */
 
} /* namespace gum */
 
#endif /* DOXYGEN_SHOULD_SKIP_THIS */