Skip to main content

Shattering Coefficient

  • Reference work entry
  • First Online:
Encyclopedia of Machine Learning and Data Mining
  • 96 Accesses

Synonyms

Growth function

Definition

The shattering coefficient \(S_{\mathcal{F}}(n)\) is a function that measures the size of a function class \(\mathcal{F}\) when its functions \(f : \mathcal{X} \rightarrow \mathbb{R}\) are restricted to sets of points \(\mathbf{x} = (x_{1},\ldots,x_{n}) \in \mathcal{X}^{n}\) of size n. Specifically, for each \(n \in \mathbb{N}\) the shattering coefficient is the maximum size of the set of vectors \(\mathcal{F}_{\mathbf{x}} =\{ (f(x_{1}),.\ldots,f(x_{n})) : f \in \mathcal{F}\}\subset \mathbb{R}^{n}\) that can be realized for some choice of \(\mathbf{x} \in \mathcal{X}^{n}\). That is,

$$\displaystyle\begin{array}{rcl} S_{\mathcal{F}}(n) =\sup \limits _{\mathbf{x}\in \mathcal{X}^{n}}\vert \mathcal{F}_{x}\vert . \end{array}$$

The shattering coefficient of a hypothesis class \(\mathcal{H}\) is used in generalization bounds as an analogue to the class’s size in the finite case.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 699.99
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Hardcover Book
USD 949.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Editor information

Editors and Affiliations

Rights and permissions

Reprints and permissions

Copyright information

© 2017 Springer Science+Business Media New York

About this entry

Cite this entry

(2017). Shattering Coefficient. In: Sammut, C., Webb, G.I. (eds) Encyclopedia of Machine Learning and Data Mining. Springer, Boston, MA. https://doi.org/10.1007/978-1-4899-7687-1_759

Download citation

Publish with us

Policies and ethics