Derivative-Free optimization algorithms. These algorithms do not require gradient information. More importantly, they can be used to solve non-smooth optimization problems.
