| | |
Summary: Learning with Distance Substitution Kernels
Bernard Haasdonk1
and Claus Bahlmann2
1
Computer Science Department
Albert-Ludwigs-University Freiburg
79110 Freiburg, Germany
haasdonk@informatik.uni-freiburg.de
2
Siemens Corporate Research, Inc.
755 College Road East
Princeton, NJ 08540, USA
claus.bahlmann@scr.siemens.com
Abstract. During recent years much effort has been spent in incorporating prob-
lem specific a-priori knowledge into kernel methods for machine learning. A
common example is a-priori knowledge given by a distance measure between
objects. A simple but effective approach for kernel construction consists of substi-
tuting the Euclidean distance in ordinary kernel functions by the problem specific
distance measure. We formalize this distance substitution procedure and investi-
gate theoretical and empirical effects. In particular we state criteria for definite-
|
