neural_net Namespace Reference
[Neural network]

Neural network namespace. More...

Data Structures

• class Basic_function

  Basic class for defining functions. More...
  

• class Basic_generalized_training_weight

  Template class for the generalized training functions. More...
  

• class Basic_neuron

  Basic_neuron template class. More...
  

• class Basic_neuron< Activation_function_type(typename Binary_operation_type::result_type), Binary_operation_type(Value_type, Value_type) >
• class Basic_topology

  Basic class for topologies. More...
  

• class Basic_training_functional

  Basic trainng functional. More...
  

• class Basic_wta_training_functional

  Class that is basic for Winner Takes All (WTA) algorithms. More...
  

• class Basic_wtm_training_functional

  Class that is basic for Winner Takes Most (WTM) algorithms. More...
  

• class Cauchy_function

  Functor that computes Cauchy hat function. More...
  

• class City_topology

  Topology for neural network that calculates distance between two neurons. More...
  

• class Classic_training_weight

  Template class for the generalize training functions that calculates generalized weight classical way. More...
  

• class Constant_function

  Functor that computes constant function. More...
  

• class Experimental_training_weight

  Template class for the generalize training functions that calculates generalizeg weight in experimental way. More...
  

• class External_randomize

  This class force not to use srand() in generation proces, therefore srand function is initialized externally. More...
  

• class Gauss_function

  Functor that compute Gauss hat function. More...
  

• class Hexagonal_topology

  Topology for neural network that calculates distance between two neurons. More...
  

• class Internal_randomize

  This class force to use srand() in generation proces, therefore srand function is NOT initialized. More...
  

• class Linear_numeric_iterator

  Class that defines bahavior of the linear numeric iterator. It begins with given value and iterate using given step. More...
  

• class Max_topology

  Topology for neural network that calculates distance between two neurons. More...
  

• class Basic_activation_function
• class Numeric_iterator

  Class that defines bahavior of the numeric iterator. More...
  

• class Randomize_policy
• class Ranges

  Class creates and claculates ranges of data. More...
  

• class Rectangular_container
• class Wta_proportional_training_functional

  Class that certain kind of WTA algorithm. More...
  

• class Wta_training_algorithm

  Class contains functionality for training kohonen network using WTA method. More...
  

• class Wtm_classical_training_functional

  Class that is basic for Winner Takes Most (WTM) algorithms. More...
  

• class Wtm_training_algorithm

  Class contains functionality for training kohonen network using WTM method. More...
  

Typedefs

• typedef Linear_numeric_iterator<
  int > linear_numeric_iterator
• typedef std::pair< size_t,
  size_t > Matrix_index

Functions

• template<typename Data_container_type, typename Kohonen_network_type, typename Randomize_policy> void generate_kohonen_network (const size_t &no_rows, const size_t &no_columns, const typename Kohonen_network_type::value_type::activation_function_type &activation_function, const typename Kohonen_network_type::value_type::binary_operation_type &binary_operation, Data_container_type &data, Kohonen_network_type &kohonen_network, const Randomize_policy &randomize_policy) throw ()
• template<typename T, typename U> std::ostream & print_network (std::ostream &os, const T &network, const U &value) throw ()
• template<typename T> std::ostream & print_network_weights (std::ostream &os, const T &network) throw ()
• template<typename T, template< typename T > class CONT> std::ostream & value_to_ostream (std::ostream &os, const CONT< T > &value) throw ()
• template<typename T> std::ostream & value_to_ostream (std::ostream &os, const T &value) throw ()

---

Detailed Description

Neural network namespace.

---

Typedef Documentation

typedef Linear_numeric_iterator< int > neural_net::linear_numeric_iterator

typedef std::pair< size_t, size_t > neural_net::Matrix_index

Stores size of container.

---

Function Documentation

template<typename Data_container_type, typename Kohonen_network_type, typename Randomize_policy> void neural_net::generate_kohonen_network (  const size_t &  no_rows, const size_t &  no_columns, const typename Kohonen_network_type::value_type::activation_function_type &  activation_function, const typename Kohonen_network_type::value_type::binary_operation_type &  binary_operation, Data_container_type &  data, Kohonen_network_type &  kohonen_network, const Randomize_policy &  randomize_policy )  throw ()

Function generates randomly distributed weights for neural network. Distribution is uniform and weights are generated based on multidimensional ranges of training data. Parameters: no_rows  is a number of rows that will be created in neural network. no_columns  is a number of columns that will be created in neural network. activation_function  is activation function that will be set. binary_operation  is a function tat will be set under activation function. data  is a reference to data container. kohonen_network  is a reference to the network. randomize_policy  is a policy class for setting up random number generator. Todo: TODO: When Rectangular_container will be changed then this class should be repaired too.

template<typename T, typename U> std::ostream& neural_net::print_network (  std::ostream &  os, const T &  network, const U &  value )  throw ()

Function prints structure and results of the neural network. Parameters: os  is output stream. network  is a reference to network. value  is a value that network will calculate results. Returns: modified stream.

template<typename T> std::ostream& neural_net::print_network_weights (  std::ostream &  os, const T &  network )  throw ()

Function prints weight of the neural network. Parameters: os  is output stream. network  is a reference to network. Returns: modified stream.

template<typename T, template< typename T > class CONT> std::ostream& neural_net::value_to_ostream (  std::ostream &  os, const CONT< T > &  value )  throw () [inline]

Function puts container of any values to the stream. Parameters: os  is a output stream. value  is a value. Returns: modified stream.

template<typename T> std::ostream& neural_net::value_to_ostream (  std::ostream &  os, const T &  value )  throw () [inline]

Function puts values to the stream. Parameters: os  is a output stream. value  is a value. Returns: modified stream.

---