Method update_mini_batch

Update the network's weights and biases by applying gradient descent using backpropagation to a single mini batch. The ``mini_batch`` is a list of tuples ``(x, y)``, and ``eta`` is the learning rate.

(self, mini_batch, eta)