ApplyConvexCombinationFactor(alfa: Double);
alfa. Transformation coefficient. The value is in the range [0.0; 1.0].
The ApplyConvexCombinationFactor method applies convex combinatorial transformation to the input elements of the network.
Convex combinatorial transformation is performed according to the formula:
Vi = alfa(t) * Vi + (1-alfa(t)) * (1/sqrt(number_of_network_inputs)
The INeuralNetwork.SetInputValuesConvex and INeuralNetwork.SetInputValuesConvexEx methods execute the given transformation.
Prior to calling ApplyConvexCombinationFactor there should be input network values assigned. To do this, use one of the methods:
Usually, value of the alfa coefficient is changed during all the process of the network learning. This makes learning more accurate.
The method use is given in the example for INeuralNetwork.SetInputValues.
See also: