softmax (Neural Network Toolbox)

Soft max transfer function

A = softmax(N)

info = softmax(code)

softmax is a transfer function. Transfer functions calculate a layer's output from its net input.

softmax(N) takes one input argument,

N - S x Q matrix of net input (column) vectors.

and returns output vectors with elements between 0 and 1, but with their size relations intact.

softmax('code') returns information about this function.

These codes are defined:

'deriv' - Name of derivative function.

'name' - Full name.

'output' - Output range.

'active' - Active input range.

compet does not have a derivative function.

Examples

Here we define a net input vector N, calculate the output, and plot both with bar graphs.

n = [0; 1; -0.5; 0.5];
a = softmax(n);
subplot(2,1,1), bar(n), ylabel('n')
subplot(2,1,2), bar(a), ylabel('a')

To change a network so that a layer uses softmax, set net.layers{i,j}.transferFcn to 'softmax'.

Call sim to simulate the network with softmax. See newc or newpnn for simulation examples.