Back to Search Start Over

[Untitled]

Authors :
Jürgen Forster
Niels Schmitt
Hans Ulrich Simon
Thorsten Suttorp
Source :
Machine Learning. 51:263-281
Publication Year :
2003
Publisher :
Springer Science and Business Media LLC, 2003.

Abstract

Concept classes can canonically be represented by matrices with entries 1 and −1. We use the singular value decomposition of this matrix to determine the optimal margins of embeddings of the concept classes of singletons and of half intervals in homogeneous Euclidean half spaces. For these concept classes the singular value decomposition can be used to construct optimal embeddings and also to prove the corresponding best possible upper bounds on the margin. We show that the optimal margin for embedding n singletons is \frac{n}{3n-4} and that the optimal margin for half intervals over l1,…,nr is \frac{\pi}{2 \ln n} + \Theta (\frac{1}{(\ln n)^2}). For the upper bounds on the margins we generalize a bound by Forster (2001). We also determine the optimal margin of some concept classes defined by circulant matrices up to a small constant factor, and we discuss the concept classes of monomials to point out limitations of our approach.

Details

ISSN :
08856125
Volume :
51
Database :
OpenAIRE
Journal :
Machine Learning
Accession number :
edsair.doi...........68b7d15a854bc6362101df825aad598f
Full Text :
https://doi.org/10.1023/a:1022905618164