The problem with marriage is that it ends every night after making love, and it must be rebuilt every morning before breakfast.

In the context of computational learning, the concept of 'probably approximately correct' (PAC) learning provides a framework for understanding the efficiency and feasibility of learning algorithms.