dsatlins (Neural Network Toolbox)

Derivative of symmetric saturating linear transfer function

dA_dN = dsatlins(N,A)

dsatlins is the derivative function for satlins.

dsatlins(N,A) takes two arguments,

N - S x Q net input.

A - S x Q output.

and returns the S x Q derivative dA/dN.

Examples

Here we define the net input N for a layer of 3 satlins neurons.

N = [0.1; 0.8; -0.7];

We calculate the layer's output A with satlins and then the derivative of A with respect to N.

A = satlins(N)
dA_dN = dsatlins(N,A)

The derivative of satlins is calculated as follows:

d = 1, if -1 <= n <= 1; 0, Otherwise.