Packing Set Systems of Finite VC dimension

Given a set of points XX together with a metric dd, an ϵ\epsilon-set N⊆XN\subseteq X is a set of point such that for every point x∈Xx\in X, there exists a point n(x)∈Nn(x)\in N such that d(x,n(x))≤ϵd(x,n(x))\leq\epsilon. It is a classical problem of discrete and computational geometry to find ϵ\epsilon-nets of small size. In the late 1980s, Haussler and Welzl showed bounds on the size of such sets in the realm of much more abstract and less geometric sets of points, using only the finiteness of their VC-dimension. Their result shows that sets of VC-dimension at most dd behave much like Euclidean balls of dimension dd.

picture

This write-up with Lucy Davidson contains an intuitive proof of the Primal Shatter Lemma (or Sauer-Shelah-Perlas Lemma) and the easiest proof I know of Haussler’s packing lemma.


(Primal Shatter Lemma) Let XX be a set of nn elements and H={H1,H2,…,Hm}\mathcal H = \{H_1, H_2, \ldots, H_m\} be a set system of VC dimension dd on XX. For all A⊆XA \subseteq X, we have

∣H∣A∣≤∑i=0d(∣A∣i)|\mathcal H_{|A}| \leq \sum_{i=0}^{d} \binom{|A|}{i}

When ∣A∣≥d|A| \geq d, the previous sum can be upper bounded by (e∣A∣d)d=O(∣A∣d)\left(\frac{e|A|}{d}\right)^d = O(|A|^d).


picture


(Haussler’s Packing Lemma) Let XX be a set of nn elements and H={H1,H2,…,Hm}\mathcal H=\{H_1, H_2, \ldots, H_m\} be a set system of VC dimension dd on XX. If H\mathcal H is ϵ\epsilon-separated then: ∣H∣=O((nϵ)d)|\mathcal H| = O\left(\left(\frac{n}{\epsilon}\right)^d\right).


✧