new RR on support vector neural networks

[Thomas Friess]

a new research report on support vector neural networks is
available at:
http://www.brunner-edv.com/friess/index.html


Support Vector Neural Networks: The Kernel Adatron with Bias 
and Soft Margin


Abstract: The kernel Adatron with bias and soft margin (KAb) is a new 
neural network alternative to support vector (SV) machines. It can learn
large-margin decision functions in kernel feature spaces in an iterative
"on-line" fashion which are identical to support vector machines.
Support vector learning is batch learning and is strongly based on solving
constrained quadratic programming problems which are nontrivial to
implement and may be subject to stability problems.
The kernel Adatron algorithm (KA), which has been developed as a joint
project, has been introduced recently. So far it has been assumed that the
bias parameter of the plane in feature space is always zero, and that all
patterns can be correctly classified by the learning machine. These
assumptions cannot always be made.
At first perceptrons in the data dependent representation, support vector
machines, and the kernel Adatron will be reviewd. Then the kernel Adatron
with bias and soft margin will be introduced.
The algorithm is conceptually simple and implements an iterative form of
unconstrained quadratic programming.
Experimental results using benchmarks and real data are provided which
allow to compare the performance and speed of kernel Adatrons and
SV machines.