FANN DatatypesThe two main datatypes used in the fann library is struct fann, which represents an artificial neural network, and struct fann_train_data, which represent training data. Summary | The two main datatypes used in the fann library is struct fann, which represents an artificial neural network, and struct fann_train_data, which represent training data. | | | | fann_type is the type used for the weights, inputs and outputs of the neural network. | | | | | | Constant array consisting of the names for the training algorithms, so that the name of an training function can be received by: | | The activation functions used for the neurons during training. | | Constant array consisting of the names for the activation function, so that the name of an activation function can be received by: | | Error function used during training. | | Constant array consisting of the names for the training error functions, so that the name of an error function can be received by: | | Stop criteria used during training. | | Constant array consisting of the names for the training stop functions, so that the name of a stop function can be received by: | | | | Constant array consisting of the names for the network types, so that the name of an network type can be received by: | | | | | | | | The fast artificial neural network(fann) structure. | | | | Describes a connection between two neurons and its weight |
fann_typefann_type is the type used for the weights, inputs and outputs of the neural network. fann_type is defined as afloat | if you include fann.h or floatfann.h | double | if you include doublefann.h | int | if you include fixedfann.h (please be aware that fixed point usage is only to be used during execution, and not during training). |
Enumerations and Constants
fann_train_enumThe Training algorithms used when training on struct fann_train_data with functions like fann_train_on_data or fann_train_on_file. The incremental training looks alters the weights after each time it is presented an input pattern, while batch only alters the weights once after it has been presented to all the patterns. FANN_TRAIN_INCREMENTAL | Standard backpropagation algorithm, where the weights are updated after each training pattern. This means that the weights are updated many times during a single epoch. For this reason some problems, will train very fast with this algorithm, while other more advanced problems will not train very well. | FANN_TRAIN_BATCH | Standard backpropagation algorithm, where the weights are updated after calculating the mean square error for the whole training set. This means that the weights are only updated once during a epoch. For this reason some problems, will train slower with this algorithm. But since the mean square error is calculated more correctly than in incremental training, some problems will reach a better solutions with this algorithm. | FANN_TRAIN_RPROP | A more advanced batch training algorithm which achieves good results for many problems. The RPROP training algorithm is adaptive, and does therefore not use the learning_rate. Some other parameters can however be set to change the way the RPROP algorithm works, but it is only recommended for users with insight in how the RPROP training algorithm works. The RPROP training algorithm is described by [Riedmiller and Braun, 1993], but the actual learning algorithm used here is the iRPROP- training algorithm which is described by [Igel and Husken, 2000] which is an variety of the standard RPROP training algorithm. | FANN_TRAIN_QUICKPROP | A more advanced batch training algorithm which achieves good results for many problems. The quickprop training algorithm uses the learning_rate parameter along with other more advanced parameters, but it is only recommended to change these advanced parameters, for users with insight in how the quickprop training algorithm works. The quickprop training algorithm is described by [Fahlman, 1988]. |
See alsofann_set_training_algorithm, fann_get_training_algorithm
FANN_TRAIN_NAMESConstant array consisting of the names for the training algorithms, so that the name of an training function can be received by: char *name = FANN_TRAIN_NAMES[train_function];
See Alsofann_train_enum
fann_activationfunc_enumThe activation functions used for the neurons during training. The activation functions can either be defined for a group of neurons by fann_set_activation_function_hidden and fann_set_activation_function_output or it can be defined for a single neuron by fann_set_activation_function. The steepness of an activation function is defined in the same way by fann_set_activation_steepness_hidden, fann_set_activation_steepness_output and fann_set_activation_steepness. The functions are described with functions where- x is the input to the activation function,
- y is the output,
- s is the steepness and
- d is the derivation.
FANN_LINEAR | Linear activation function. |
- span: -inf < y < inf
- y = x*s, d = 1*s
- Can NOT be used in fixed point.
FANN_THRESHOLD | Threshold activation function. |
- x < 0 -> y = 0, x >= 0 -> y = 1
- Can NOT be used during training.
FANN_THRESHOLD_SYMMETRIC | Threshold activation function. |
- x < 0 -> y = 0, x >= 0 -> y = 1
- Can NOT be used during training.
FANN_SIGMOID | Sigmoid activation function. |
- One of the most used activation functions.
- span: 0 < y < 1
- y = 1/(1 + exp(-2*s*x))
- d = 2*s*y*(1 - y)
FANN_SIGMOID_STEPWISE | Stepwise linear approximation to sigmoid. |
- Faster than sigmoid but a bit less precise.
FANN_SIGMOID_SYMMETRIC | Symmetric sigmoid activation function, aka. tanh. |
- One of the most used activation functions.
- span: -1 < y < 1
- y = tanh(s*x) = 2/(1 + exp(-2*s*x)) - 1
- d = s*(1-(y*y))
FANN_SIGMOID_SYMMETRIC | Stepwise linear approximation to symmetric sigmoid. |
- Faster than symmetric sigmoid but a bit less precise.
FANN_GAUSSIAN | Gaussian activation function. |
- 0 when x = -inf, 1 when x = 0 and 0 when x = inf
- span: 0 < y < 1
- y = exp(-x*s*x*s)
- d = -2*x*s*y*s
FANN_GAUSSIAN_SYMMETRIC | Symmetric gaussian activation function. |
- -1 when x = -inf, 1 when x = 0 and 0 when x = inf
- span: -1 < y < 1
- y = exp(-x*s*x*s)*2-1
- d = -2*x*s*(y+1)*s
FANN_ELLIOT | Fast (sigmoid like) activation function defined by David Elliott |
- span: 0 < y < 1
- y = ((x*s) / 2) / (1 + |x*s|) + 0.5
- d = s*1/(2*(1+|x*s|)*(1+|x*s|))
FANN_ELLIOT_SYMMETRIC | Fast (symmetric sigmoid like) activation function defined by David Elliott |
- span: -1 < y < 1
- y = (x*s) / (1 + |x*s|)
- d = s*1/((1+|x*s|)*(1+|x*s|))
FANN_LINEAR_PIECE | Bounded linear activation function. |
- span: 0 <= y <= 1
- y = x*s, d = 1*s
FANN_LINEAR_PIECE_SYMMETRIC | Bounded linear activation function. |
- span: -1 <= y <= 1
- y = x*s, d = 1*s
FANN_SIN_SYMMETRIC | Periodical sinus activation function. |
- span: -1 <= y <= 1
- y = sin(x*s)
- d = s*cos(x*s)
FANN_COS_SYMMETRIC | Periodical cosinus activation function. |
- span: -1 <= y <= 1
- y = cos(x*s)
- d = s*-sin(x*s)
See alsofann_set_activation_function_hidden, fann_set_activation_function_output
FANN_ACTIVATIONFUNC_NAMESConstant array consisting of the names for the activation function, so that the name of an activation function can be received by: char *name = FANN_ACTIVATIONFUNC_NAMES[activation_function];
See Alsofann_activationfunc_enum
fann_errorfunc_enumError function used during training. FANN_ERRORFUNC_LINEAR | Standard linear error function. | FANN_ERRORFUNC_TANH | Tanh error function, usually better but can require a lower learning rate. This error function agressively targets outputs that differ much from the desired, while not targetting outputs that only differ a little that much. This activation function is not recommended for cascade training and incremental training. |
See alsofann_set_train_error_function, fann_get_train_error_function
FANN_ERRORFUNC_NAMESConstant array consisting of the names for the training error functions, so that the name of an error function can be received by: char *name = FANN_ERRORFUNC_NAMES[error_function];
See Alsofann_errorfunc_enum
FANN_STOPFUNC_NAMESConstant array consisting of the names for the training stop functions, so that the name of a stop function can be received by: char *name = FANN_STOPFUNC_NAMES[stop_function];
See Alsofann_stopfunc_enum
fann_network_type_enumDefinition of network types used by fann_get_network_type FANN_NETTYPE_LAYER | Each layer only has connections to the next layer | FANN_NETTYPE_SHORTCUT | Each layer has connections to all following layers |
See Alsofann_get_network_type This enumeration appears in FANN >= 2.1.0
FANN_NETWORK_TYPE_NAMESConstant array consisting of the names for the network types, so that the name of an network type can be received by: char *network_type_name = FANN_NETWORK_TYPE_NAMES[fann_get_network_type(ann)];
See Alsofann_get_network_type This constant appears in FANN >= 2.1.0
fann_callback_typeThis callback function can be called during training when using fann_train_on_data, fann_train_on_file or fann_cascadetrain_on_data. typedef int (FANN_API * fann_callback_type) (struct fann *ann, struct fann_train_data *train, unsigned int max_epochs, unsigned int epochs_between_reports, float desired_error, unsigned int epochs);
The callback can be set by using fann_set_callback and is very usefull for doing custom things during training. It is recommended to use this function when implementing custom training procedures, or when visualizing the training in a GUI etc. The parameters which the callback function takes is the parameters given to the fann_train_on_data, plus an epochs parameter which tells how many epochs the training have taken so far. The callback function should return an integer, if the callback function returns -1, the training will terminate. Example of a callback functionint FANN_API test_callback(struct fann *ann, struct fann_train_data *train, unsigned int max_epochs, unsigned int epochs_between_reports, float desired_error, unsigned int epochs) { printf("Epochs %8d. MSE: %.5f. Desired-MSE: %.5f\n", epochs, fann_get_MSE(ann), desired_error); return 0; }
See alsofann_set_callback, fann_train_on_data
struct fannThe fast artificial neural network(fann) structure. Data within this structure should never be accessed directly, but only by using the fann_get_... and fann_set_... functions. The fann structure is created using one of the fann_create_... functions and each of the functions which operates on the structure takes struct fann * ann as the first parameter. See alsofann_create_standard, fann_destroy Summary | | | Describes a connection between two neurons and its weight |
fann_connectionDescribes a connection between two neurons and its weight from_neuron | Unique number used to identify source neuron | to_neuron | Unique number used to identify destination neuron | weight | The numerical value of the weight |
See Alsofann_get_connection_array, fann_set_weight_array This structure appears in FANN >= 2.1.0
|