SANN - Automated Network Search (ANS) - MLP Activation Functions Tab

Select the MLP activation functions tab of the SANN - Automated Network Search (ANS) dialog box to select the type of activation functions that are to be included in the Automated Network Search (ANS) for both hidden (input-hidden) and output (hidden-output) units for multilayer perceptron networks. For information on the options that are common to all tabs, see STATISTICA Automated Network Search (ANS).

Note: For RBF neural networks, the hidden activation functions are always set to isotropic Gaussian basis functions. For regression analysis, the only activation available for the outputs units of an RBF are linear functions. For classification tasks with cross-entropy error function, the output activation function of RBF and MLP networks are set to softmax.

Hidden neurons. This group box provides a list of hidden activation functions to choose from for inclusion in the Automated Network Search (ANS). Note that you can select more than one activation function at a time:

Identity. Uses the identity function. With this function, the activation level is passed on directly as the output of the neurons.

Logistic. Uses the logistic sigmoid function. This is an S-shaped (sigmoid) curve, with output in the range (0,1).

Tanh. Uses the hyperbolic tangent function (recommended). The hyperbolic tangent function (tanh) is a symmetric S-shaped (sigmoid) function, whose output lies in the range (-1, +1). Often performs better than the logistic sigmoid function because of its symmetry.

Exp. Uses the negative exponential activation function.

Sine. Uses the standard sine activation function.

Output neurons. This group box provides a list of output activation functions to choose from for inclusion in the Automated Network Search (ANS). Note that you can select more than one activation function at a time:

Identity. Uses the identity function (recommended). With this function, the activation level is passed on directly as the output of the neurons. This is the only activation function available for RBF networks when the error function is SOS.

Logistic. Uses the logistic sigmoid function. This is an S-shaped (sigmoid) curve, with output in the range (0, 1).

Tanh. Uses the hyperbolic tangent function. The hyperbolic tangent function (tanh) is a symmetric S-shaped (sigmoid) function, whose output lies in the range (-1, +1). Often performs better than the logistic sigmoid function because of its symmetry.

Exponential. Uses the negative exponential activation function.

Sine. Uses the standard sine activation function.