Learn how SGD speeds up training by updating weights per sample. Compare Batch, Stochastic, and Mini-batch gradient descent.