Scale-sensitive Dimensions, Uniform Convergence,

Learnability in Valiant’s PAC learning model has been shown to be strongly related to the existence of uniform laws of large numbers. These laws define a distribution-free convergence property of means to expectations uniformly over classes of random variables. Classes of real-valued functions enjoying such a properly are also known as uniform Glivenko-Cantelli classes. In this paper we prove, through a generalization of Sauer’s lemma that may be interesting in its own righi, a new characterization of uniform Glivenko-Cantelli classes. Our characterization yields Dudley, Gint!, and Zinn’s previous characterization as a corollary. Furthermore, it is the first based on a simple combinatorial quantify generalizing the VapnikChervonenkis dimension. We apply this result to characterize PAC leamability in the statistical regression framework of probabilistic concepts, solving an open problem posed by Keams and Schapire. Our characterization shows that the accuracy parameter plays a crucial role in determining the effective complexity of the learner’s hypothesis class.