PAC-learnable if and only if finite VC dimension

machine-learning

Statement

Lemma

A hypothesis space $H$ is PAC learnable if and only if the VC dimension is finite.

Proof