Solution

ID: past-exam-of-the-computer-science-course-of-the-university-of-oxford/2015/computational-learning-theory/1/b/solution

Fix an -point set . The restriction of an intersection to is determined by the ordered pair of restrictions . Some pairs can yield the same intersection, so the growth function of a hypothesis class satisfies
Assume first that . If points were shattered by , the Sauer-Shelah lemma would imply
But , so . Raising this strict inequality to the positive integer power gives a contradiction. No -point set can therefore be shattered; neither can any larger set, since shattering passes to subsets. Consequently
The printed strict inequality needs . For example, on a nonempty domain let contain only the identically zero classifier. Then , whereas the stated conclusion would require . More generally, a nonempty hypothesis class of VC dimension zero has only one distinct labelling function, since two different functions would shatter a point on which they differ. Its self-intersection class also has VC dimension zero.

New to topics? Read the docs here!