Approximation as the fitting of a target by a hypothesis. Bias.This is the accompanying discussion for Noiseless Learning.