gradient-descent

ml.doc (1.2): Learning as optimization

Finding a good classifier by stumbling in its general direction.