Implement Momentum and Regularization #4
Description
In an effort to make the standard feedforward networks more useful, we need to implement momentum and regularization (with an arbitrary regularization function -- l1, l2), which should improve learning.