fminunc isnot Gradient Descent but a Advanced Optimization

The Matlab/Octave fminunc function uses a variant of the BFGS method that was mentioned in the Coursera lectures. If you would like to know more about it, google for  Broyden-Fletcher-Goldfarb-Shanno algorithm. But you do not need to know the details. Most cases built-in functions in software packages hide all the unnecessary details of the implemented very sophisticated and advanced numerical methods. You only have to know what the algorithm does, and how to call it, which is usually well described in the help.