Fixed topology training	The size and topology of the ANN is determined in advance and the training alters the weights in order to minimize the difference between the desired output values and the actual output values. This kind of training is supported by fann_train_on_data.
Evolving topology training	The training start out with an empty ANN, only consisting of input and output neurons. Hidden neurons and connections is the added during training, in order to reach the same goal as for fixed topology training. This kind of training is supported by FANN Cascade Training.

Summary

FANN Training	There are many different ways of training neural networks and the FANN library supports a number of different approaches.
struct fann_train_data	Structure used to store data, for use with training.
FANN Training
Training
fann_train	Train one iteration with a set of inputs, and a set of desired outputs.
fann_test	Test with a set of inputs, and a set of desired outputs.
fann_get_MSE	Reads the mean square error from the network.
fann_get_bit_fail	The number of fail bits; means the number of output neurons which differ more than the bit fail limit (see fann_get_bit_fail_limit, fann_set_bit_fail_limit).
fann_reset_MSE	Resets the mean square error from the network.
Training Data Training
fann_train_on_data	Trains on an entire dataset, for a period of time.
fann_train_on_file	Does the same as fann_train_on_data, but reads the training data directly from a file.
fann_train_epoch	Train one epoch with a set of training data.
fann_test_data	Test a set of training data and calculates the MSE for the training data.
Training Data Manipulation
fann_read_train_from_file	Reads a file that stores training data.
fann_create_train	Creates an empty training data struct.
fann_create_train_from_callback	Creates the training data struct from a user supplied function.
fann_destroy_train	Destructs the training data and properly deallocates all of the associated data.
fann_shuffle_train_data	Shuffles training data, randomizing the order.
fann_scale_train	Scale input and output data based on previously calculated parameters.
fann_descale_train	Descale input and output data based on previously calculated parameters.
fann_set_input_scaling_params	Calculate input scaling parameters for future use based on training data.
fann_set_output_scaling_params	Calculate output scaling parameters for future use based on training data.
fann_set_scaling_params	Calculate input and output scaling parameters for future use based on training data.
fann_clear_scaling_params	Clears scaling parameters.
fann_scale_input	Scale data in input vector before feed it to ann based on previously calculated parameters.
fann_scale_output	Scale data in output vector before feed it to ann based on previously calculated parameters.
fann_descale_input	Scale data in input vector after get it from ann based on previously calculated parameters.
fann_descale_output	Scale data in output vector after get it from ann based on previously calculated parameters.
fann_scale_input_train_data	Scales the inputs in the training data to the specified range.
fann_scale_output_train_data	Scales the outputs in the training data to the specified range.
fann_scale_train_data	Scales the inputs and outputs in the training data to the specified range.
fann_merge_train_data	Merges the data from data1 and data2 into a new struct fann_train_data.
fann_duplicate_train_data	Returns an exact copy of a struct fann_train_data.
fann_subset_train_data	Returns an copy of a subset of the struct fann_train_data, starting at position pos and length elements forward.
fann_length_train_data	Returns the number of training patterns in the struct fann_train_data.
fann_num_input_train_data	Returns the number of inputs in each of the training patterns in the struct fann_train_data.
fann_num_output_train_data	Returns the number of outputs in each of the training patterns in the struct fann_train_data.
fann_save_train	Save the training structure to a file, with the format as specified in fann_read_train_from_file
fann_save_train_to_fixed	Saves the training structure to a fixed point data file.
Parameters
fann_get_training_algorithm	Return the training algorithm as described by fann_train_enum.
fann_set_training_algorithm	Set the training algorithm.
fann_get_learning_rate	Return the learning rate.
fann_set_learning_rate	Set the learning rate.
fann_get_learning_momentum	Get the learning momentum.
fann_set_learning_momentum	Set the learning momentum.
fann_get_activation_function	Get the activation function for neuron number neuron in layer number layer, counting the input layer as layer 0.
fann_set_activation_function	Set the activation function for neuron number neuron in layer number layer, counting the input layer as layer 0.
fann_set_activation_function_layer	Set the activation function for all the neurons in the layer number layer, counting the input layer as layer 0.
fann_set_activation_function_hidden	Set the activation function for all of the hidden layers.
fann_set_activation_function_output	Set the activation function for the output layer.
fann_get_activation_steepness	Get the activation steepness for neuron number neuron in layer number layer, counting the input layer as layer 0.
fann_set_activation_steepness	Set the activation steepness for neuron number neuron in layer number layer, counting the input layer as layer 0.
fann_set_activation_steepness_layer	Set the activation steepness all of the neurons in layer number layer, counting the input layer as layer 0.
fann_set_activation_steepness_hidden	Set the steepness of the activation steepness in all of the hidden layers.
fann_set_activation_steepness_output	Set the steepness of the activation steepness in the output layer.
fann_get_train_error_function	Returns the error function used during training.
fann_set_train_error_function	Set the error function used during training.
fann_get_train_stop_function	Returns the the stop function used during training.
fann_set_train_stop_function	Set the stop function used during training.
fann_get_bit_fail_limit	Returns the bit fail limit used during training.
fann_set_bit_fail_limit	Set the bit fail limit used during training.
fann_set_callback	Sets the callback function for use during training.
fann_get_quickprop_decay	The decay is a small negative valued number which is the factor that the weights should become smaller in each iteration during quickprop training.
fann_set_quickprop_decay	Sets the quickprop decay factor.
fann_get_quickprop_mu	The mu factor is used to increase and decrease the step-size during quickprop training.
fann_set_quickprop_mu	Sets the quickprop mu factor.
fann_get_rprop_increase_factor	The increase factor is a value larger than 1, which is used to increase the step-size during RPROP training.
fann_set_rprop_increase_factor	The increase factor used during RPROP training.
fann_get_rprop_decrease_factor	The decrease factor is a value smaller than 1, which is used to decrease the step-size during RPROP training.
fann_set_rprop_decrease_factor	The decrease factor is a value smaller than 1, which is used to decrease the step-size during RPROP training.
fann_get_rprop_delta_min	The minimum step-size is a small positive number determining how small the minimum step-size may be.
fann_set_rprop_delta_min	The minimum step-size is a small positive number determining how small the minimum step-size may be.
fann_get_rprop_delta_max	The maximum step-size is a positive number determining how large the maximum step-size may be.
fann_set_rprop_delta_max	The maximum step-size is a positive number determining how large the maximum step-size may be.
fann_get_rprop_delta_zero	The initial step-size is a positive number determining the initial step size.
fann_set_rprop_delta_zero	The initial step-size is a positive number determining the initial step size.
fann_get_sarprop_weight_decay_shift	The sarprop weight decay shift.
fann_set_sarprop_weight_decay_shift	Set the sarprop weight decay shift.
fann_get_sarprop_step_error_threshold_factor	The sarprop step error threshold factor.
fann_set_sarprop_step_error_threshold_factor	Set the sarprop step error threshold factor.
fann_get_sarprop_step_error_shift	The get sarprop step error shift.
fann_set_sarprop_step_error_shift	Set the sarprop step error shift.
fann_get_sarprop_temperature	The sarprop weight decay shift.
fann_set_sarprop_temperature	Set the sarprop_temperature.

struct fann_train_data

Structure used to store data, for use with training.

The data inside this structure should never be manipulated directly, but should use some of the supplied functions in Training Data Manipulation.

The training data structure is very usefull for storing data during training and testing of a neural network.

FANN Training

Summary

Training
fann_train	Train one iteration with a set of inputs, and a set of desired outputs.
fann_test	Test with a set of inputs, and a set of desired outputs.
fann_get_MSE	Reads the mean square error from the network.
fann_get_bit_fail	The number of fail bits; means the number of output neurons which differ more than the bit fail limit (see fann_get_bit_fail_limit, fann_set_bit_fail_limit).
fann_reset_MSE	Resets the mean square error from the network.
Training Data Training
fann_train_on_data	Trains on an entire dataset, for a period of time.
fann_train_on_file	Does the same as fann_train_on_data, but reads the training data directly from a file.
fann_train_epoch	Train one epoch with a set of training data.
fann_test_data	Test a set of training data and calculates the MSE for the training data.
Training Data Manipulation
fann_read_train_from_file	Reads a file that stores training data.
fann_create_train	Creates an empty training data struct.
fann_create_train_from_callback	Creates the training data struct from a user supplied function.
fann_destroy_train	Destructs the training data and properly deallocates all of the associated data.
fann_shuffle_train_data	Shuffles training data, randomizing the order.
fann_scale_train	Scale input and output data based on previously calculated parameters.
fann_descale_train	Descale input and output data based on previously calculated parameters.
fann_set_input_scaling_params	Calculate input scaling parameters for future use based on training data.
fann_set_output_scaling_params	Calculate output scaling parameters for future use based on training data.
fann_set_scaling_params	Calculate input and output scaling parameters for future use based on training data.
fann_clear_scaling_params	Clears scaling parameters.
fann_scale_input	Scale data in input vector before feed it to ann based on previously calculated parameters.
fann_scale_output	Scale data in output vector before feed it to ann based on previously calculated parameters.
fann_descale_input	Scale data in input vector after get it from ann based on previously calculated parameters.
fann_descale_output	Scale data in output vector after get it from ann based on previously calculated parameters.
fann_scale_input_train_data	Scales the inputs in the training data to the specified range.
fann_scale_output_train_data	Scales the outputs in the training data to the specified range.
fann_scale_train_data	Scales the inputs and outputs in the training data to the specified range.
fann_merge_train_data	Merges the data from data1 and data2 into a new struct fann_train_data.
fann_duplicate_train_data	Returns an exact copy of a struct fann_train_data.
fann_subset_train_data	Returns an copy of a subset of the struct fann_train_data, starting at position pos and length elements forward.
fann_length_train_data	Returns the number of training patterns in the struct fann_train_data.
fann_num_input_train_data	Returns the number of inputs in each of the training patterns in the struct fann_train_data.
fann_num_output_train_data	Returns the number of outputs in each of the training patterns in the struct fann_train_data.
fann_save_train	Save the training structure to a file, with the format as specified in fann_read_train_from_file
fann_save_train_to_fixed	Saves the training structure to a fixed point data file.
Parameters
fann_get_training_algorithm	Return the training algorithm as described by fann_train_enum.
fann_set_training_algorithm	Set the training algorithm.
fann_get_learning_rate	Return the learning rate.
fann_set_learning_rate	Set the learning rate.
fann_get_learning_momentum	Get the learning momentum.
fann_set_learning_momentum	Set the learning momentum.
fann_get_activation_function	Get the activation function for neuron number neuron in layer number layer, counting the input layer as layer 0.
fann_set_activation_function	Set the activation function for neuron number neuron in layer number layer, counting the input layer as layer 0.
fann_set_activation_function_layer	Set the activation function for all the neurons in the layer number layer, counting the input layer as layer 0.
fann_set_activation_function_hidden	Set the activation function for all of the hidden layers.
fann_set_activation_function_output	Set the activation function for the output layer.
fann_get_activation_steepness	Get the activation steepness for neuron number neuron in layer number layer, counting the input layer as layer 0.
fann_set_activation_steepness	Set the activation steepness for neuron number neuron in layer number layer, counting the input layer as layer 0.
fann_set_activation_steepness_layer	Set the activation steepness all of the neurons in layer number layer, counting the input layer as layer 0.
fann_set_activation_steepness_hidden	Set the steepness of the activation steepness in all of the hidden layers.
fann_set_activation_steepness_output	Set the steepness of the activation steepness in the output layer.
fann_get_train_error_function	Returns the error function used during training.
fann_set_train_error_function	Set the error function used during training.
fann_get_train_stop_function	Returns the the stop function used during training.
fann_set_train_stop_function	Set the stop function used during training.
fann_get_bit_fail_limit	Returns the bit fail limit used during training.
fann_set_bit_fail_limit	Set the bit fail limit used during training.
fann_set_callback	Sets the callback function for use during training.
fann_get_quickprop_decay	The decay is a small negative valued number which is the factor that the weights should become smaller in each iteration during quickprop training.
fann_set_quickprop_decay	Sets the quickprop decay factor.
fann_get_quickprop_mu	The mu factor is used to increase and decrease the step-size during quickprop training.
fann_set_quickprop_mu	Sets the quickprop mu factor.
fann_get_rprop_increase_factor	The increase factor is a value larger than 1, which is used to increase the step-size during RPROP training.
fann_set_rprop_increase_factor	The increase factor used during RPROP training.
fann_get_rprop_decrease_factor	The decrease factor is a value smaller than 1, which is used to decrease the step-size during RPROP training.
fann_set_rprop_decrease_factor	The decrease factor is a value smaller than 1, which is used to decrease the step-size during RPROP training.
fann_get_rprop_delta_min	The minimum step-size is a small positive number determining how small the minimum step-size may be.
fann_set_rprop_delta_min	The minimum step-size is a small positive number determining how small the minimum step-size may be.
fann_get_rprop_delta_max	The maximum step-size is a positive number determining how large the maximum step-size may be.
fann_set_rprop_delta_max	The maximum step-size is a positive number determining how large the maximum step-size may be.
fann_get_rprop_delta_zero	The initial step-size is a positive number determining the initial step size.
fann_set_rprop_delta_zero	The initial step-size is a positive number determining the initial step size.
fann_get_sarprop_weight_decay_shift	The sarprop weight decay shift.
fann_set_sarprop_weight_decay_shift	Set the sarprop weight decay shift.
fann_get_sarprop_step_error_threshold_factor	The sarprop step error threshold factor.
fann_set_sarprop_step_error_threshold_factor	Set the sarprop step error threshold factor.
fann_get_sarprop_step_error_shift	The get sarprop step error shift.
fann_set_sarprop_step_error_shift	Set the sarprop step error shift.
fann_get_sarprop_temperature	The sarprop weight decay shift.
fann_set_sarprop_temperature	Set the sarprop_temperature.

Training

fann_train

FANN_EXTERNAL void FANN_API fann_train( struct fann * ann,
fann_type * input,
fann_type * desired_output )

Train one iteration with a set of inputs, and a set of desired outputs. This training is always incremental training (see fann_train_enum), since only one pattern is presented.

Parameters

ann	The neural network structure
input	an array of inputs. This array must be exactly fann_get_num_input long.
desired_output	an array of desired outputs. This array must be exactly fann_get_num_output long.

fann_test

FANN_EXTERNAL fann_type * FANN_API fann_test( struct fann * ann,
fann_type * input,
fann_type * desired_output )

Test with a set of inputs, and a set of desired outputs. This operation updates the mean square error, but does not change the network in any way.

fann_get_MSE

FANN_EXTERNAL float FANN_API fann_get_MSE( struct fann * ann )

Reads the mean square error from the network.

Reads the mean square error from the network. This value is calculated during training or testing, and can therefore sometimes be a bit off if the weights have been changed since the last calculation of the value.

fann_get_bit_fail

FANN_EXTERNAL unsigned int FANN_API fann_get_bit_fail( struct fann * ann )

The number of fail bits; means the number of output neurons which differ more than the bit fail limit (see fann_get_bit_fail_limit, fann_set_bit_fail_limit). The bits are counted in all of the training data, so this number can be higher than the number of training data.

This value is reset by fann_reset_MSE and updated by all the same functions which also updates the MSE value (e.g. fann_test_data, fann_train_epoch)

fann_reset_MSE

FANN_EXTERNAL void FANN_API fann_reset_MSE( struct fann * ann )

Resets the mean square error from the network.

This function also resets the number of bits that fail.

Training Data Training

fann_train_on_data

FANN_EXTERNAL void FANN_API fann_train_on_data(
    struct fann * ann,
    struct fann_train_data * data,
    unsigned int max_epochs,
    unsigned int epochs_between_reports,
    float desired_error
)

Trains on an entire dataset, for a period of time.

This training uses the training algorithm chosen by fann_set_training_algorithm, and the parameters set for these training algorithms.

Parameters

ann	The neural network
data	The data, which should be used during training
max_epochs	The maximum number of epochs the training should continue
epochs_between_reports	The number of epochs between printing a status report to stdout. A value of zero means no reports should be printed.
desired_error	The desired fann_get_MSE or fann_get_bit_fail, depending on which stop function is chosen by fann_set_train_stop_function.

Instead of printing out reports every epochs_between_reports, a callback function can be called (see fann_set_callback).

fann_train_on_file

FANN_EXTERNAL void FANN_API fann_train_on_file(
    struct fann * ann,
    const char * filename,
    unsigned int max_epochs,
    unsigned int epochs_between_reports,
    float desired_error
)

Does the same as fann_train_on_data, but reads the training data directly from a file.

fann_train_epoch

FANN_EXTERNAL float FANN_API fann_train_epoch( struct fann * ann,
struct fann_train_data * data )

Train one epoch with a set of training data.

Train one epoch with the training data stored in data. One epoch is where all of the training data is considered exactly once.

This function returns the MSE error as it is calculated either before or during the actual training. This is not the actual MSE after the training epoch, but since calculating this will require to go through the entire training set once more, it is more than adequate to use this value during training.

The training algorithm used by this function is chosen by the fann_set_training_algorithm function.

fann_test_data

FANN_EXTERNAL float FANN_API fann_test_data( struct fann * ann,
struct fann_train_data * data )

Test a set of training data and calculates the MSE for the training data.

This function updates the MSE and the bit fail values.

Training Data Manipulation

fann_read_train_from_file

FANN_EXTERNAL struct fann_train_data *FANN_API fann_read_train_from_file(
const char * filename
)

Reads a file that stores training data.

The file must be formatted like

num_train_data num_input num_output
inputdata seperated by space
outputdata seperated by space

.
.
.

inputdata seperated by space
outputdata seperated by space

fann_create_train

FANN_EXTERNAL struct fann_train_data * FANN_API fann_create_train(
    unsigned int num_data,
    unsigned int num_input,
    unsigned int num_output
)

Creates an empty training data struct.

fann_create_train_from_callback

FANN_EXTERNAL struct fann_train_data * FANN_API fann_create_train_from_callback(
    unsigned int num_data,
    unsigned int num_input,
    unsigned int num_output,
    void (FANN_API *user_function)( unsigned int, unsigned int, unsigned int, fann_type * , fann_type * )
)

Creates the training data struct from a user supplied function. As the training data are numerable (data 1, data 2...), the user must write a function that receives the number of the training data set (input,output) and returns the set.

Parameters

num_data	The number of training data
num_input	The number of inputs per training data
num_output	The number of ouputs per training data
user_function	The user suplied function

Parameters for the user function

num	The number of the training data set
num_input	The number of inputs per training data
num_output	The number of ouputs per training data
input	The set of inputs
output	The set of desired outputs

fann_destroy_train

FANN_EXTERNAL void FANN_API fann_destroy_train(
struct fann_train_data * train_data
)

Destructs the training data and properly deallocates all of the associated data. Be sure to call this function after finished using the training data.

This function appears in FANN >= 1.0.0

fann_shuffle_train_data

FANN_EXTERNAL void FANN_API fann_shuffle_train_data(
struct fann_train_data * train_data
)

Shuffles training data, randomizing the order. This is recommended for incremental training, while it have no influence during batch training.

This function appears in FANN >= 1.1.0.

fann_scale_train

FANN_EXTERNAL void FANN_API fann_scale_train( struct fann * ann,
struct fann_train_data * data )

Scale input and output data based on previously calculated parameters.

Parameters

ann	ann for which were calculated trained parameters before
data	training data that needs to be scaled

fann_descale_train

FANN_EXTERNAL void FANN_API fann_descale_train( struct fann * ann,
struct fann_train_data * data )

Descale input and output data based on previously calculated parameters.

Parameters

ann	ann for which were calculated trained parameters before
data	training data that needs to be descaled

fann_set_input_scaling_params

FANN_EXTERNAL int FANN_API fann_set_input_scaling_params(
    struct fann * ann,
    const struct fann_train_data * data,
    float new_input_min,
    float new_input_max
)

Calculate input scaling parameters for future use based on training data.

Parameters

ann	ann for wgich parameters needs to be calculated
data	training data that will be used to calculate scaling parameters
new_input_min	desired lower bound in input data after scaling (not strictly followed)
new_input_max	desired upper bound in input data after scaling (not strictly followed)

fann_set_output_scaling_params

FANN_EXTERNAL int FANN_API fann_set_output_scaling_params(
    struct fann * ann,
    const struct fann_train_data * data,
    float new_output_min,
    float new_output_max
)

Calculate output scaling parameters for future use based on training data.

Parameters

ann	ann for wgich parameters needs to be calculated
data	training data that will be used to calculate scaling parameters
new_output_min	desired lower bound in input data after scaling (not strictly followed)
new_output_max	desired upper bound in input data after scaling (not strictly followed)

fann_set_scaling_params

FANN_EXTERNAL int FANN_API fann_set_scaling_params(
    struct fann * ann,
    const struct fann_train_data * data,
    float new_input_min,
    float new_input_max,
    float new_output_min,
    float new_output_max
)

Calculate input and output scaling parameters for future use based on training data.

Parameters

ann	ann for wgich parameters needs to be calculated
data	training data that will be used to calculate scaling parameters
new_input_min	desired lower bound in input data after scaling (not strictly followed)
new_input_max	desired upper bound in input data after scaling (not strictly followed)
new_output_min	desired lower bound in input data after scaling (not strictly followed)
new_output_max	desired upper bound in input data after scaling (not strictly followed)

fann_clear_scaling_params

FANN_EXTERNAL int FANN_API fann_clear_scaling_params( struct fann * ann )

Clears scaling parameters.

Parameters

ann	ann for which to clear scaling parameters

This function appears in FANN >= 2.1.0

fann_scale_input

FANN_EXTERNAL void FANN_API fann_scale_input( struct fann * ann,
fann_type * input_vector )

Scale data in input vector before feed it to ann based on previously calculated parameters.

Parameters

ann	for which scaling parameters were calculated
input_vector	input vector that will be scaled

fann_scale_output

FANN_EXTERNAL void FANN_API fann_scale_output( struct fann * ann,
fann_type * output_vector )

Scale data in output vector before feed it to ann based on previously calculated parameters.

Parameters

ann	for which scaling parameters were calculated
output_vector	output vector that will be scaled

fann_descale_input

FANN_EXTERNAL void FANN_API fann_descale_input( struct fann * ann,
fann_type * input_vector )

Scale data in input vector after get it from ann based on previously calculated parameters.

Parameters

ann	for which scaling parameters were calculated
input_vector	input vector that will be descaled

fann_descale_output

FANN_EXTERNAL void FANN_API fann_descale_output( struct fann * ann,
fann_type * output_vector )

Scale data in output vector after get it from ann based on previously calculated parameters.

Parameters

ann	for which scaling parameters were calculated
output_vector	output vector that will be descaled

fann_scale_input_train_data

FANN_EXTERNAL void FANN_API fann_scale_input_train_data(
    struct fann_train_data * train_data,
    fann_type new_min,
    fann_type new_max
)

Scales the inputs in the training data to the specified range.

fann_scale_output_train_data

FANN_EXTERNAL void FANN_API fann_scale_output_train_data(
    struct fann_train_data * train_data,
    fann_type new_min,
    fann_type new_max
)

Scales the outputs in the training data to the specified range.

fann_scale_train_data

FANN_EXTERNAL void FANN_API fann_scale_train_data(
    struct fann_train_data * train_data,
    fann_type new_min,
    fann_type new_max
)

Scales the inputs and outputs in the training data to the specified range.

fann_merge_train_data

FANN_EXTERNAL struct fann_train_data *FANN_API fann_merge_train_data(
struct fann_train_data * data1,
struct fann_train_data * data2
)

Merges the data from data1 and data2 into a new struct fann_train_data.

This function appears in FANN >= 1.1.0.

fann_duplicate_train_data

FANN_EXTERNAL struct fann_train_data *FANN_API fann_duplicate_train_data(
struct fann_train_data * data
)

Returns an exact copy of a struct fann_train_data.

This function appears in FANN >= 1.1.0.

fann_subset_train_data

FANN_EXTERNAL struct fann_train_data *FANN_API fann_subset_train_data(
    struct fann_train_data * data,
    unsigned int pos,
    unsigned int length
)

Returns an copy of a subset of the struct fann_train_data, starting at position pos and length elements forward.

fann_subset_train_data(train_data, 0, fann_length_train_data(train_data))

Will do the same as fann_duplicate_train_data.

fann_length_train_data

FANN_EXTERNAL unsigned int FANN_API fann_length_train_data(
struct fann_train_data * data
)

Returns the number of training patterns in the struct fann_train_data.

This function appears in FANN >= 2.0.0.

fann_num_input_train_data

FANN_EXTERNAL unsigned int FANN_API fann_num_input_train_data(
struct fann_train_data * data
)

Returns the number of inputs in each of the training patterns in the struct fann_train_data.

fann_num_output_train_data

FANN_EXTERNAL unsigned int FANN_API fann_num_output_train_data(
struct fann_train_data * data
)

Returns the number of outputs in each of the training patterns in the struct fann_train_data.

fann_save_train

FANN_EXTERNAL int FANN_API fann_save_train( struct fann_train_data * data,
const char * filename )

Save the training structure to a file, with the format as specified in fann_read_train_from_file

Return

The function returns 0 on success and -1 on failure.

fann_save_train_to_fixed

FANN_EXTERNAL int FANN_API fann_save_train_to_fixed(
    struct fann_train_data * data,
    const char * filename,
    unsigned int decimal_point
)

Saves the training structure to a fixed point data file.

This function is very usefull for testing the quality of a fixed point network.

Return

The function returns 0 on success and -1 on failure.

Parameters

fann_get_training_algorithm

FANN_EXTERNAL enum fann_train_enum FANN_API fann_get_training_algorithm(
struct fann * ann
)

Return the training algorithm as described by fann_train_enum. This training algorithm is used by fann_train_on_data and associated functions.

Note that this algorithm is also used during fann_cascadetrain_on_data, although only FANN_TRAIN_RPROP and FANN_TRAIN_QUICKPROP is allowed during cascade training.

The default training algorithm is FANN_TRAIN_RPROP.

fann_set_training_algorithm

FANN_EXTERNAL void FANN_API fann_set_training_algorithm(
struct fann * ann,
enum fann_train_enum training_algorithm
)

Set the training algorithm.

More info available in fann_get_training_algorithm

This function appears in FANN >= 1.0.0.

fann_get_learning_rate

FANN_EXTERNAL float FANN_API fann_get_learning_rate( struct fann * ann )

Return the learning rate.

The learning rate is used to determine how aggressive training should be for some of the training algorithms (FANN_TRAIN_INCREMENTAL, FANN_TRAIN_BATCH, FANN_TRAIN_QUICKPROP). Do however note that it is not used in FANN_TRAIN_RPROP.

The default learning rate is 0.7.

fann_set_learning_rate

FANN_EXTERNAL void FANN_API fann_set_learning_rate( struct fann * ann,
float learning_rate )

Set the learning rate.

More info available in fann_get_learning_rate

This function appears in FANN >= 1.0.0.

fann_get_learning_momentum

FANN_EXTERNAL float FANN_API fann_get_learning_momentum( struct fann * ann )

Get the learning momentum.

The learning momentum can be used to speed up FANN_TRAIN_INCREMENTAL training. A too high momentum will however not benefit training. Setting momentum to 0 will be the same as not using the momentum parameter. The recommended value of this parameter is between 0.0 and 1.0.

The default momentum is 0.

fann_set_learning_momentum

FANN_EXTERNAL void FANN_API fann_set_learning_momentum( struct fann * ann,
float learning_momentum )

Set the learning momentum.

More info available in fann_get_learning_momentum

This function appears in FANN >= 2.0.0.

fann_get_activation_function

FANN_EXTERNAL enum fann_activationfunc_enum FANN_API fann_get_activation_function(
    struct fann * ann,
    int layer,
    int neuron
)

Get the activation function for neuron number neuron in layer number layer, counting the input layer as layer 0.

It is not possible to get activation functions for the neurons in the input layer.

Information about the individual activation functions is available at fann_activationfunc_enum.

Returns

The activation function for the neuron or -1 if the neuron is not defined in the neural network.

fann_set_activation_function

FANN_EXTERNAL void FANN_API fann_set_activation_function(
    struct fann * ann,
    enum fann_activationfunc_enum activation_function,
    int layer,
    int neuron
)

Set the activation function for neuron number neuron in layer number layer, counting the input layer as layer 0.

It is not possible to set activation functions for the neurons in the input layer.

When choosing an activation function it is important to note that the activation functions have different range. FANN_SIGMOID is e.g. in the 0 - 1 range while FANN_SIGMOID_SYMMETRIC is in the -1 - 1 range and FANN_LINEAR is unbound.

Information about the individual activation functions is available at fann_activationfunc_enum.

The default activation function is FANN_SIGMOID_STEPWISE.

fann_set_activation_function_layer

FANN_EXTERNAL void FANN_API fann_set_activation_function_layer(
    struct fann * ann,
    enum fann_activationfunc_enum activation_function,
    int layer
)

Set the activation function for all the neurons in the layer number layer, counting the input layer as layer 0.

It is not possible to set activation functions for the neurons in the input layer.

fann_set_activation_function_hidden

FANN_EXTERNAL void FANN_API fann_set_activation_function_hidden(
struct fann * ann,
enum fann_activationfunc_enum activation_function
)

Set the activation function for all of the hidden layers.

fann_set_activation_function_output

FANN_EXTERNAL void FANN_API fann_set_activation_function_output(
struct fann * ann,
enum fann_activationfunc_enum activation_function
)

Set the activation function for the output layer.

fann_get_activation_steepness

FANN_EXTERNAL fann_type FANN_API fann_get_activation_steepness(
    struct fann * ann,
    int layer,
    int neuron
)

Get the activation steepness for neuron number neuron in layer number layer, counting the input layer as layer 0.

It is not possible to get activation steepness for the neurons in the input layer.

The steepness of an activation function says something about how fast the activation function goes from the minimum to the maximum. A high value for the activation function will also give a more agressive training.

When training neural networks where the output values should be at the extremes (usually 0 and 1, depending on the activation function), a steep activation function can be used (e.g. 1.0).

The default activation steepness is 0.5.

Returns

The activation steepness for the neuron or -1 if the neuron is not defined in the neural network.

fann_set_activation_steepness

FANN_EXTERNAL void FANN_API fann_set_activation_steepness( struct fann * ann,
fann_type steepness,
int layer,
int neuron )

Set the activation steepness for neuron number neuron in layer number layer, counting the input layer as layer 0.

It is not possible to set activation steepness for the neurons in the input layer.

When training neural networks where the output values should be at the extremes (usually 0 and 1, depending on the activation function), a steep activation function can be used (e.g. 1.0).

The default activation steepness is 0.5.

fann_set_activation_steepness_layer

FANN_EXTERNAL void FANN_API fann_set_activation_steepness_layer(
    struct fann * ann,
    fann_type steepness,
    int layer
)

Set the activation steepness all of the neurons in layer number layer, counting the input layer as layer 0.

It is not possible to set activation steepness for the neurons in the input layer.

fann_set_activation_steepness_hidden

FANN_EXTERNAL void FANN_API fann_set_activation_steepness_hidden(
struct fann * ann,
fann_type steepness
)

Set the steepness of the activation steepness in all of the hidden layers.

fann_set_activation_steepness_output

FANN_EXTERNAL void FANN_API fann_set_activation_steepness_output(
struct fann * ann,
fann_type steepness
)

Set the steepness of the activation steepness in the output layer.

fann_get_train_error_function

FANN_EXTERNAL enum fann_errorfunc_enum FANN_API fann_get_train_error_function(
struct fann * ann
)

Returns the error function used during training.

The error functions is described further in fann_errorfunc_enum

The default error function is FANN_ERRORFUNC_TANH

fann_set_train_error_function

FANN_EXTERNAL void FANN_API fann_set_train_error_function(
struct fann * ann,
enum fann_errorfunc_enum train_error_function
)

Set the error function used during training.

The error functions is described further in fann_errorfunc_enum

fann_get_train_stop_function

FANN_EXTERNAL enum fann_stopfunc_enum FANN_API fann_get_train_stop_function(
struct fann * ann
)

Returns the the stop function used during training.

The stop function is described further in fann_stopfunc_enum

The default stop function is FANN_STOPFUNC_MSE

fann_set_train_stop_function

FANN_EXTERNAL void FANN_API fann_set_train_stop_function(
struct fann * ann,
enum fann_stopfunc_enum train_stop_function
)

Set the stop function used during training.

Returns the the stop function used during training.

The stop function is described further in fann_stopfunc_enum

fann_get_bit_fail_limit

FANN_EXTERNAL fann_type FANN_API fann_get_bit_fail_limit( struct fann * ann )

Returns the bit fail limit used during training.

The bit fail limit is used during training where the fann_stopfunc_enum is set to FANN_STOPFUNC_BIT.

The limit is the maximum accepted difference between the desired output and the actual output during training. Each output that diverges more than this limit is counted as an error bit. This difference is divided by two when dealing with symmetric activation functions, so that symmetric and not symmetric activation functions can use the same limit.

The default bit fail limit is 0.35.

fann_set_bit_fail_limit

FANN_EXTERNAL void FANN_API fann_set_bit_fail_limit( struct fann * ann,
fann_type bit_fail_limit )

Set the bit fail limit used during training.

fann_set_callback

FANN_EXTERNAL void FANN_API fann_set_callback( struct fann * ann,
fann_callback_type callback )

Sets the callback function for use during training.

See fann_callback_type for more information about the callback function.

The default callback function simply prints out some status information.

This function appears in FANN >= 2.0.0.

fann_get_quickprop_decay

FANN_EXTERNAL float FANN_API fann_get_quickprop_decay( struct fann * ann )

The decay is a small negative valued number which is the factor that the weights should become smaller in each iteration during quickprop training. This is used to make sure that the weights do not become too high during training.

The default decay is -0.0001.

fann_set_quickprop_decay

FANN_EXTERNAL void FANN_API fann_set_quickprop_decay( struct fann * ann,
float quickprop_decay )

Sets the quickprop decay factor.

fann_get_quickprop_mu

FANN_EXTERNAL float FANN_API fann_get_quickprop_mu( struct fann * ann )

The mu factor is used to increase and decrease the step-size during quickprop training. The mu factor should always be above 1, since it would otherwise decrease the step-size when it was suppose to increase it.

The default mu factor is 1.75.

fann_set_quickprop_mu

FANN_EXTERNAL void FANN_API fann_set_quickprop_mu( struct fann * ann,
float quickprop_mu )

Sets the quickprop mu factor.

fann_get_rprop_increase_factor

FANN_EXTERNAL float FANN_API fann_get_rprop_increase_factor( struct fann * ann )

The increase factor is a value larger than 1, which is used to increase the step-size during RPROP training.

The default increase factor is 1.2.

fann_set_rprop_increase_factor

FANN_EXTERNAL void FANN_API fann_set_rprop_increase_factor(
struct fann * ann,
float rprop_increase_factor
)

The increase factor used during RPROP training.

fann_get_rprop_decrease_factor

FANN_EXTERNAL float FANN_API fann_get_rprop_decrease_factor( struct fann * ann )

The decrease factor is a value smaller than 1, which is used to decrease the step-size during RPROP training.

The default decrease factor is 0.5.

fann_set_rprop_decrease_factor

FANN_EXTERNAL void FANN_API fann_set_rprop_decrease_factor(
struct fann * ann,
float rprop_decrease_factor
)

The decrease factor is a value smaller than 1, which is used to decrease the step-size during RPROP training.

fann_get_rprop_delta_min

FANN_EXTERNAL float FANN_API fann_get_rprop_delta_min( struct fann * ann )

The minimum step-size is a small positive number determining how small the minimum step-size may be.

The default value delta min is 0.0.

fann_set_rprop_delta_min

FANN_EXTERNAL void FANN_API fann_set_rprop_delta_min( struct fann * ann,
float rprop_delta_min )

The minimum step-size is a small positive number determining how small the minimum step-size may be.

fann_get_rprop_delta_max

FANN_EXTERNAL float FANN_API fann_get_rprop_delta_max( struct fann * ann )

The maximum step-size is a positive number determining how large the maximum step-size may be.

The default delta max is 50.0.

fann_set_rprop_delta_max

FANN_EXTERNAL void FANN_API fann_set_rprop_delta_max( struct fann * ann,
float rprop_delta_max )

The maximum step-size is a positive number determining how large the maximum step-size may be.

fann_get_rprop_delta_zero

FANN_EXTERNAL float FANN_API fann_get_rprop_delta_zero( struct fann * ann )

The initial step-size is a positive number determining the initial step size.

The default delta zero is 0.1.

fann_set_rprop_delta_zero

FANN_EXTERNAL void FANN_API fann_set_rprop_delta_zero( struct fann * ann,
float rprop_delta_max )

The initial step-size is a positive number determining the initial step size.

fann_get_sarprop_weight_decay_shift

FANN_EXTERNAL float FANN_API fann_get_sarprop_weight_decay_shift(
struct fann * ann
)

The sarprop weight decay shift.

The default delta max is -6.644.

fann_set_sarprop_weight_decay_shift

FANN_EXTERNAL void FANN_API fann_set_sarprop_weight_decay_shift(
struct fann * ann,
float sarprop_weight_decay_shift
)

Set the sarprop weight decay shift.

This function appears in FANN >= 2.1.0.

fann_get_sarprop_step_error_threshold_factor

FANN_EXTERNAL float FANN_API fann_get_sarprop_step_error_threshold_factor(
struct fann * ann
)

The sarprop step error threshold factor.

The default delta max is 0.1.

fann_set_sarprop_step_error_threshold_factor

FANN_EXTERNAL void FANN_API fann_set_sarprop_step_error_threshold_factor(
struct fann * ann,
float sarprop_step_error_threshold_factor
)

Set the sarprop step error threshold factor.

This function appears in FANN >= 2.1.0.

fann_get_sarprop_step_error_shift

FANN_EXTERNAL float FANN_API fann_get_sarprop_step_error_shift(
struct fann * ann
)

The get sarprop step error shift.

The default delta max is 1.385.

fann_set_sarprop_step_error_shift

FANN_EXTERNAL void FANN_API fann_set_sarprop_step_error_shift(
struct fann * ann,
float sarprop_step_error_shift
)

Set the sarprop step error shift.

This function appears in FANN >= 2.1.0.

fann_get_sarprop_temperature

FANN_EXTERNAL float FANN_API fann_get_sarprop_temperature( struct fann * ann )

The sarprop weight decay shift.

The default delta max is 0.015.

fann_set_sarprop_temperature

FANN_EXTERNAL void FANN_API fann_set_sarprop_temperature(
struct fann * ann,
float sarprop_temperature
)

Set the sarprop_temperature.

This function appears in FANN >= 2.1.0.

FANN Training

Two fundementally different approaches are the most commonly used

struct fann_train_data

See also

FANN Training

Training

fann_train

Parameters

See also

fann_test

See also

fann_get_MSE

See also

fann_get_bit_fail

See also

fann_reset_MSE

See also

Training Data Training

fann_train_on_data

Parameters

See also

fann_train_on_file

See also

fann_train_epoch

See also

fann_test_data

See also

Training Data Manipulation

fann_read_train_from_file

The file must be formatted like

See also

fann_create_train

See also

fann_create_train_from_callback

Parameters

Parameters for the user function

See also

fann_destroy_train

fann_shuffle_train_data

fann_scale_train

Parameters

See also

fann_descale_train

Parameters

See also

fann_set_input_scaling_params

Parameters

See also

fann_set_output_scaling_params

Parameters

See also

fann_set_scaling_params

Parameters

See also

fann_clear_scaling_params

Parameters

fann_scale_input

Parameters

See also

fann_scale_output

Parameters

See also

fann_descale_input

Parameters

See also

fann_descale_output

Parameters

See also

fann_scale_input_train_data

See also

fann_scale_output_train_data

See also

fann_scale_train_data

See also

fann_merge_train_data

fann_duplicate_train_data

fann_subset_train_data

See also

fann_length_train_data

fann_num_input_train_data