Gradient-based adaptation of general gaussian kernels

Tobias Glasmachers, Christian Igel

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningpeer review

51 Citationer (Scopus)

Abstract

Gradient-based optimizing of gaussian kernel functions is considered. The gradient for the adaptation of scaling and rotation of the input space is computed to achieve invariance against linear transformations. This is done by using the exponential map as a parameterization of the kernel parameter manifold. By restricting the optimization to a constant trace subspace, the kernel size can be controlled. This is, for example, useful to prevent overfitting when minimizing radius-margin generalization performance measures. The concepts are demonstrated by training hard margin support vector machines on toy data.
OriginalsprogEngelsk
TidsskriftNeural Computation
Vol/bind17
Udgave nummer10
Sider (fra-til)2099-2105
Antal sider7
ISSN0899-7667
DOI
StatusUdgivet - 2005
Udgivet eksterntJa

Citationsformater