dnetsum (Neural Network Toolbox)

Sum net input derivative function

dN_dZ = dnetsum(Z,N)

dnetsum is the net input derivative function for netsum.

dnetsum takes two arguments,

Z - S x Q weighted input.

N - S x Q net input.

and returns the S x Q derivative dN/dZ.

Examples

Here we define two weighted inputs for a layer with three neurons.

Z1 = [0; 1; -1];
Z2 = [1; 0.5; 1.2];

We calculate the layer's net input N with netsum and then the derivative of N with respect to each weighted input.

N = netsum(Z1,Z2)
dN_dZ1 = dnetsum(Z1,N)
dN_dZ2 = dnetsum(Z2,N)

The derivative of a sum with respect to any element of that sum is always a ones matrix that is the same size as the sum.