probabilistic_layer.h

//   OpenNN: Open Neural Networks Library
//   www.opennn.net
//
//   P R O B A B I L I S T I C   L A Y E R   C L A S S   H E A D E R
//
//   Artificial Intelligence Techniques SL
//   artelnics@artelnics.com
 
#ifndef PROBABILISTICLAYER_H
#define PROBABILISTICLAYER_H
 
// System includes
 
#include <cmath>
#include <cstdlib>
#include <fstream>
#include <iostream>
#include <string>
#include <sstream>
 
// OpenNN includes
 
#include "config.h"
 
#include "layer.h"
 
#include "opennn_strings.h"
 
 
namespace OpenNN
{
 
struct ProbabilisticLayerForwardPropagation;
struct ProbabilisticLayerBackPropagation;
struct ProbabilisticLayerBackPropagationLM;
 
#ifdef OPENNN_CUDA
    #include "../../opennn-cuda/opennn-cuda/struct_probabilistic_layer_cuda.h"
#endif
 
 
 
 
class ProbabilisticLayer : public Layer
{
 
public:
 
   // Constructors
 
   explicit ProbabilisticLayer();
 
   explicit ProbabilisticLayer(const Index&, const Index&);
 
   // Destructor
 
   virtual ~ProbabilisticLayer();
 
   // Enumerations
 
 
   enum class ActivationFunction{Binary, Logistic, Competitive, Softmax};
 
   // Get methods
 
   Index get_inputs_number() const;
   Index get_neurons_number() const;
 
   Index get_biases_number() const;
   Index get_synaptic_weights_number() const;
 
   const type& get_decision_threshold() const;
 
   const ActivationFunction& get_activation_function() const;
   string write_activation_function() const;
   string write_activation_function_text() const;
 
   const bool& get_display() const;
 
   // Set methods
 
   void set();
   void set(const Index&, const Index&);
   void set(const ProbabilisticLayer&);
 
   void set_inputs_number(const Index&);
   void set_neurons_number(const Index&);
 
   void set_biases(const Tensor<type, 2>&);
   void set_synaptic_weights(const Tensor<type, 2>&);
 
   void set_parameters(const Tensor<type, 1>&, const Index& index=0);
   void set_decision_threshold(const type&);
 
   void set_activation_function(const ActivationFunction&);
   void set_activation_function(const string&);
 
   virtual void set_default();
 
   // Parameters
 
   const Tensor<type, 2>& get_biases() const;
   const Tensor<type, 2>& get_synaptic_weights() const;
 
   Tensor<type, 2> get_biases(Tensor<type, 1>&) const;
   Tensor<type, 2> get_synaptic_weights(Tensor<type, 1>&) const;
 
   Index get_parameters_number() const;
   Tensor<type, 1> get_parameters() const;
 
   // Display messages
 
   void set_display(const bool&);
 
   // Parameters initialization methods
 
   void set_biases_constant(const type&);
   void set_synaptic_weights_constant(const type&);
   void set_synaptic_weights_constant_Glorot();
 
   void set_parameters_constant(const type&);
 
   void set_parameters_random();
 
   void insert_parameters(const Tensor<type, 1>&, const Index&);
 
   // Combinations
 
   void calculate_combinations(const Tensor<type, 2>&,
                               const Tensor<type, 2>&,
                               const Tensor<type, 2>&,
                               Tensor<type, 2>&) const;
 
   // Activations
 
   void calculate_activations(const Tensor<type, 2>&, Tensor<type, 2>&) const;
 
   void calculate_activations_derivatives(const Tensor<type, 2>&,
                                          Tensor<type, 2>&,
                                          Tensor<type, 3>&) const;
 
   // Outputs
 
   Tensor<type, 2> calculate_outputs(const Tensor<type, 2>&);
 
   void forward_propagate(const Tensor<type, 2>&,
                          LayerForwardPropagation*);
 
   void forward_propagate(const Tensor<type, 2>&,
                          Tensor<type, 1>,
                          LayerForwardPropagation*);
 
   // Gradient methods
 
   void calculate_error_gradient(const Tensor<type, 2>&,
                                 LayerForwardPropagation*,
                                 LayerBackPropagation*) const;
 
   void insert_gradient(LayerBackPropagation*, const Index&, Tensor<type, 1>&) const;
 
   // Squared errors methods
 
   void calculate_squared_errors_Jacobian_lm(const Tensor<type, 2>&,
                                             LayerForwardPropagation*,
                                             LayerBackPropagationLM*);
 
   void insert_squared_errors_Jacobian_lm(LayerBackPropagationLM*,
                                          const Index&,
                                          Tensor<type, 2>&) const;
 
   // Expression methods
 
   string write_binary_expression(const Tensor<string, 1>&, const Tensor<string, 1>&) const;
   string write_logistic_expression(const Tensor<string, 1>&, const Tensor<string, 1>&) const;
   string write_competitive_expression(const Tensor<string, 1>&, const Tensor<string, 1>&) const;
   string write_softmax_expression(const Tensor<string, 1>&, const Tensor<string, 1>&) const;
   string write_no_probabilistic_expression(const Tensor<string, 1>&, const Tensor<string, 1>&) const;
 
   string write_expression(const Tensor<string, 1>&, const Tensor<string, 1>&) const;
   string write_combinations(const Tensor<string, 1>&) const;
   string write_activations(const Tensor<string, 1>&) const;
 
   string write_expression_c() const;
   string write_combinations_c() const;
   string write_activations_c() const;
 
   string write_expression_python() const;
   string write_combinations_python() const;
   string write_activations_python() const;
 
   // Serialization methods
 
   void from_XML(const tinyxml2::XMLDocument&);
 
   void write_XML(tinyxml2::XMLPrinter&) const;
 
 
protected:
 
 
   Tensor<type, 2> biases;
 
 
   Tensor<type, 2> synaptic_weights;
 
 
   ActivationFunction activation_function = ActivationFunction::Logistic;
 
   type decision_threshold;
 
 
   bool display = true;
 
#ifdef OPENNN_CUDA
    #include "../../opennn-cuda/opennn-cuda/probabilistic_layer_cuda.h"
#else
};
#endif
 
struct ProbabilisticLayerForwardPropagation : LayerForwardPropagation
{
    // Constructor
 
    explicit ProbabilisticLayerForwardPropagation() : LayerForwardPropagation()
    {
    }
 
    // Constructor
 
    explicit ProbabilisticLayerForwardPropagation(const Index& new_batch_samples_number, Layer* new_layer_pointer)
        : LayerForwardPropagation()
    {
        set(new_batch_samples_number, new_layer_pointer);
    }
 
    void set(const Index& new_batch_samples_number, Layer* new_layer_pointer)
    {
        layer_pointer = new_layer_pointer;
 
        batch_samples_number = new_batch_samples_number;
 
        const Index neurons_number = layer_pointer->get_neurons_number();
 
        combinations.resize(batch_samples_number, neurons_number);
 
        activations.resize(batch_samples_number, neurons_number);
 
        activations_derivatives.resize(batch_samples_number, neurons_number, neurons_number);
    }
 
 
    void print() const
    {
        cout << "Combinations:" << endl;
        cout << combinations << endl;
 
        cout << "Activations:" << endl;
        cout << activations << endl;
 
        cout << "Activations derivatives:" << endl;
        cout << activations_derivatives << endl;
    }
 
    Tensor<type, 2> combinations;
    Tensor<type, 2> activations;
    Tensor<type, 3> activations_derivatives;
};
 
 
struct ProbabilisticLayerBackPropagationLM : LayerBackPropagationLM
{
    explicit ProbabilisticLayerBackPropagationLM() : LayerBackPropagationLM()
    {
 
    }
 
 
    explicit ProbabilisticLayerBackPropagationLM(const Index& new_batch_samples_number, Layer* new_layer_pointer)
        : LayerBackPropagationLM()
    {
        set(new_batch_samples_number, new_layer_pointer);
    }
 
 
    void set(const Index& new_batch_samples_number, Layer* new_layer_pointer)
    {
        layer_pointer = new_layer_pointer;
 
        batch_samples_number = new_batch_samples_number;
 
        const Index neurons_number = layer_pointer->get_neurons_number();
        const Index parameters_number = layer_pointer->get_parameters_number();
 
        delta.resize(batch_samples_number, neurons_number);
        delta_row.resize(neurons_number);
 
        squared_errors_Jacobian.resize(batch_samples_number, parameters_number);
 
        error_combinations_derivatives.resize(batch_samples_number, neurons_number);
    }
 
    void print() const
    {
        cout << "Delta:" << endl;
        cout << delta << endl;
 
        cout << "Squared errors Jacobian: " << endl;
        cout << squared_errors_Jacobian << endl;
    }
 
    Tensor<type, 2> delta;
    Tensor<type, 1> delta_row;
 
    Tensor<type, 2> error_combinations_derivatives;
 
    Tensor<type, 2> squared_errors_Jacobian;
};
 
 
 
struct ProbabilisticLayerBackPropagation : LayerBackPropagation
{
    explicit ProbabilisticLayerBackPropagation() : LayerBackPropagation()
    {
 
    }
 
 
    explicit ProbabilisticLayerBackPropagation(const Index& new_batch_samples_number, Layer* new_layer_pointer)
        : LayerBackPropagation()
    {
        set(new_batch_samples_number, new_layer_pointer);
    }
 
 
    void set(const Index& new_batch_samples_number, Layer* new_layer_pointer)
    {
        layer_pointer = new_layer_pointer;
 
        batch_samples_number = new_batch_samples_number;
 
        const Index neurons_number = layer_pointer->get_neurons_number();
        const Index inputs_number = layer_pointer->get_inputs_number();
 
        biases_derivatives.resize(neurons_number);
 
        synaptic_weights_derivatives.resize(inputs_number, neurons_number);
 
        delta.resize(batch_samples_number, neurons_number);
        delta_row.resize(neurons_number);
 
        error_combinations_derivatives.resize(batch_samples_number, neurons_number);
    }
 
    void print() const
    {
        cout << "Delta:" << endl;
        cout << delta << endl;
 
        cout << "Biases derivatives:" << endl;
        cout << biases_derivatives << endl;
 
        cout << "Synaptic weights derivatives:" << endl;
        cout << synaptic_weights_derivatives << endl;
    }
 
    Tensor<type, 2> delta;
    Tensor<type, 1> delta_row;
 
    Tensor<type, 2> error_combinations_derivatives;
 
    Tensor<type, 2> synaptic_weights_derivatives;
    Tensor<type, 1> biases_derivatives;
};
 
}
 
#endif
 
 
// OpenNN: Open Neural Networks Library.
// Copyright(C) 2005-2021 Artificial Intelligence Techniques, SL.
//
// This library is free software; you can redistribute it and/or
// modify it under the terms of the GNU Lesser General Public
// License as published by the Free Software Foundation; either
// version 2.1 of the License, or any later version.
//
// This library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
// Lesser General Public License for more details.
 
// You should have received a copy of the GNU Lesser General Public
// License along with this library; if not, write to the Free Software
 
// Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA