MATLAB: Want to plot MSE for a given error

dspmse

Please somebody tell me how to write code for Mean Square error

Assume x is one vector and y is another. you can think of x as your original and y the fit.

x = randn(100,1);
y = randn(100,1);
mean(abs(x-y).^2)

If X and Y are images

   X= randn(256,256);
   Y = randn(256,256);
   absD = abs(X-Y).^2;
   sum(absD(:))/numel(X)

This is common behavior. The type of training that uses a validation set is called "Early Stopping" and "Stopped Training".

Training stops when the validation error increases a specified number of times. I think the default number is 6. However, see the documentation

help train

doc train

Also, see the discussion in the comp.ai.neural-nets FAQ.

Hope this helps.

Greg

You can use fimplicit()

fun = @(x, y) y.*sin(2*x) - x.*cos(2*y);
fimplicit(fun, [-10 10 -10 10])