- 158 Downloads
Let (M, ρ) be a metric space, let S ⊆ M, and let ε > 0. A set E ⊆ M is an ε-cover for S, if for every s ∈ S there is an e ∈ E such that ϱ(s, e) ≤ ε.
An ε-coverE is said to be proper, if E ⊆ S.
The notion of an ε-cover is frequently used in kernel-based learning methods.
For further information, we refer the reader to Herbrich (2002).
- Herbrich, R. (2002). Learning kernel classifiers: Theory and algorithms. Cambridge, MA: MIT Press.Google Scholar