Batch Gradient Descent

This performs gradient descent with the entire dataset.

The equations and their meanings are as defined in gradient descent

Pros:

1. Guaranteed to converge to global minima for convex $J$.
2. Converges to local minima for other surface types.

Cons:

1. Doesn’t work for datasets needing to be out-of-core.
2. Thus, no online learning
3. Runs slow