How do I display the connection weights after each epoch for a perceptron network using MATLAB code?

I need to know how the weights changes after each epoch. How do I display the connection weights?

Answer 1

[Mark Hudson Beale]

The function PLOTWB displays weights and biases graphically.

plotwb(net)

You can attach this function to any network object, and watch the weights change as training progresses as in this example:

[x,t] = house_dataset;
net = feedforwardnet(10);
net.plotFcns = [net.plotFcns {'plotwb'}];
net = train(net,x,t);  % Click the "Hinton Weight Bias" plot button while training.

There isn't a standard way to display numerical values during training, but you can create a custom version of the training function your network uses. I.e. make a custom version of whatever function "net.trainFcn" points to which displays weight and bias values at the command line each epoch and then set "net.trainFcn" to point to your custom function.