hrb4fann

Harbour bindings for FANN - Fast Artificial Neural Networks 2.2.0.

Preface

Harbour is a modern programming language, primarily used to create database/business programs. It is a modernized, open sourced and cross-platform version of the older and largely DOS-only Clipper system, which in turn developed from the dBase database market of the 1980s and 90s. See more info at http://www.kresin.ru/en/harbour.html

Fast Artificial Neural Network Library is a free open source neural network library, which implements multilayer artificial neural networks in C with support for both fully connected and sparsely connected networks. FANN official site is http://leenissen.dk/

Installation

First of all you need to download FANN library, unpack and compile it, using the C compiler of your choice - the same, which you use with Harbour. There are compile scripts in hrb4fann/fann directory:

• lib_fann.sh - for Linux
• lib_fann_mingw.bat - for Windows, Mingw C compiler

Before using one of them, you may need to edit it for to set the correct path to the unpacked FANN directory ( FANN_DIR variable ) and, if you work in Windows, the path to the C compiler. In case of successful compiling you'll find the library in hrb4fann/lib directory.

The next step is the compiling of hrb4funn itself with one of scripts in hrb4fann/ directory:

• lib_hrb4fann.sh - for Linux
• lib_hrb4fann_mingw.bat - for Windows, Mingw C compiler

As in case of compiling the FANN library, you need to set correct paths to the unpacked FANN directory and, for Windows, the path to the C compiler. Additionally, you need to set the path to the Harbour in HRB_DIR variable.

Functions list

• pAnn = fann_create_standard( num_layers, { num_input, ..., num_output } )

• pAnn = fann_create_sparse( connection_rate, num_layers, { num_input, ..., num_output } )

• pAnn = fann_copy( pAnn )

• fann_destroy( pAnn )

• aOutput = fann_run( pAnn, aInput )

• fann_randomize_weights( pAnn, min_weight, max_weight )

• fann_init_weights( pAnn, pdata )

• pAnn = fann_create_from_file( pAnn )

• fann_save( pAnn, cFileName )

• fann_get_num_input( pAnn )

• fann_get_num_output( pAnn )

• fann_get_total_neurons( pAnn )

• fann_get_total_connections( pAnn )

• fann_get_network_type( pAnn )

• fann_get_connection_rate( pAnn )

• fann_get_num_layers( pAnn )

• fann_get_layer_array( pAnn )

• fann_get_bias_array( pAnn )

• fann_get_connection_array( pAnn )

• fann_set_weight( pAnn )

• fann_train( pAnn, aInputs, aDesired_outputs )

• fann_test( pAnn, aInputs, aDesired_outputs )

• fann_get_MSE( pAnn )

• fann_get_bit_fail( pAnn )

• fann_reset_MSE( pAnn )

• fann_train_on_data( pAnn, pData, max_epochs, epochs_between_reports, desired_error )

• fann_train_on_file( pAnn, cFileName, max_epochs, epochs_between_reports, desired_error )

• fann_train_epoch( pAnn, pData )

• fann_test_data( pAnn, pData )

• fann_read_train_from_file( cFileName )

• fann_create_train( num_data, num_input, num_output )

• fann_destroy_train( pData )

• fann_get_input_train_data( pData, num_input )

• fann_get_output_train_data( pData, num_input )

• fann_set_train_data( pData, num, pInput, pOutput )

• fann_length_train_data( pData )

• fann_num_input_train_data( pData )

• fann_num_output_train_data( pData )

• fann_save_train( pData, cFileName )

• fann_get_training_algorithm( pAnn )

• fann_set_training_algorithm( pAnn )

• fann_get_learning_rate( pAnn )

• fann_set_learning_rate( pAnn )

• fann_get_learning_momentum( pAnn )

• fann_set_learning_momentum( pAnn )

• fann_get_activation_function( pAnn, ilayer, iNeuron )

• fann_set_activation_function( pAnn, iType, ilayer, iNeuron )

• fann_set_activation_function_layer( pAnn, iType, ilayer )

• fann_set_activation_function_hidden( pAnn, iType )

• fann_set_activation_function_output( pAnn, iType )

• fann_get_activation_steepness( pAnn, ilayer, iNeuron )

• fann_set_activation_steepness( pAnn, dSteepness, ilayer, iNeuron )

• fann_set_activation_steepness_layer( pAnn, dSteepness, ilayer )

• fann_set_activation_steepness_hidden( pAnn, dSteepness )

• fann_set_activation_steepness_output( pAnn, dSteepness )

• fann_get_train_error_function( pAnn )

• fann_set_train_error_function( pAnn, iType )

• fann_get_train_stop_function( pAnn )

• fann_set_train_stop_function( pAnn, iType )

• fann_get_bit_fail_limit( pAnn )

• fann_set_bit_fail_limit( pAnn, dLimit )

• fann_set_callback( pAnn, cFuncName )

• fann_get_quickprop_decay( pAnn )

• fann_set_quickprop_decay( pAnn, dDecay )

• fann_get_quickprop_mu( pAnn )

• fann_set_quickprop_mu( pAnn, dMU )

• fann_get_rprop_increase_factor( pAnn )

• fann_set_rprop_increase_factor( pAnn, dFactor )

• fann_get_rprop_decrease_factor( pAnn )

• fann_set_rprop_decrease_factor( pAnn, dFactor )

• fann_get_rprop_delta_min( pAnn )

• fann_set_rprop_delta_min( pAnn, ddelta_min )

• fann_get_rprop_delta_max( pAnn )

• fann_set_rprop_delta_max( pAnn, ddelta_max )

• fann_get_rprop_delta_zero( pAnn )

• fann_set_rprop_delta_zero( pAnn, ddelta_zero )

• fann_get_sarprop_weight_decay_shift( pAnn )

• fann_set_sarprop_weight_decay_shift( pAnn, dShift )

• fann_get_sarprop_step_error_threshold_factor( pAnn )

• fann_set_sarprop_step_error_threshold_factor( pAnn, dFactor )

• fann_get_sarprop_step_error_shift( pAnn )

• fann_set_sarprop_step_error_shift( pAnn, dShift )

• fann_get_sarprop_temperature( pAnn )

• fann_get_sarprop_temperature( pAnn )

• fann_set_sarprop_temperature( pAnn, dTemp )