MATLAB, supervised learning using gradient descent rule
[w,y_opt,E,niter] = grad_descent_fit(f_type, data) is a function that returns parameters 'w' for the selected function type to fit the given data. It uses gradient descent method to learn. The first column in the data file must represent input (X) and the second column (Y) must represent the corresponding output
f_ype:
1 - 2nd order polynomial
2 - 3rd order polynomial
3 - 2nd order Fourier series
4 - 4th order Fourier series
Output Arguments:
w parameters of the resultant function
y_out output generated from the resultant function
E error in each itteration
niter total number of itterations