"A key insight in computational learning theory is that the complexity of a learning problem is determined by the complexity of the hypothesis space and the amount of data available."