In practice, the total objective often includes a regularisation term that penalises model complexity:
