Generic author design sample pages 1998/07/09 14:59
11 Making Large-Scale SVM Learning Practical
Thorsten Joachims
Universitat Dortmund, Informatik, AI-Unit
Thorsten Joachims@cs.uni-dortmund.de
http://www-ai.cs.uni-dortmund.de/PERSONAL/joachims.html
Tobe published in: 'Advances in Kernel Methods - Support Vector Learning',
Bernhard Scholkopf, Christopher J. C. Burges, and Alexander J. Smola (eds.),
MIT Press, Cambridge, USA, 1998.
Training a support vector machine (SVM) leads to a quadratic optimization
problem with b ound constraints and one linear equality constraint. Despite the
fact that this type of problem is well understoo d, there are many issues to b e
considered in designing an SVM learner. In particular, for large learning tasks with
many training examples, o-the-shelf optimization techniques for general quadratic
programs quickly b ecome intractable in their memory and time requirements.
SV M
light
1
is an implementation of an SVM learner which addresses the problem of
large tasks. This chapter presents algorithmic and computational results developed
for
SV M
light
V2.0, which make large-scale SVM training more practical. The results
give guidelines for the application of SVMs to large domains.
11.1 Introduction
Chapter 1 and Vapnik (1995) showhow training a support vector machine for the
pattern recognition problem leads to the following quadratic optimization problem
(QP) OP1.
(OP1)
minimize:
W
(
)=
,
`
X
i
=1
i
+
1
2
`
X
i
=1
`
X
j
=1
y
i
y
j
i
j
k
(
x
i
;
x
j
)
(11.1)
sub ject to:
`
X
i
=1
y
i
i
=0
(11.2)
8
i
:0
i
C
(11.3)
1.
SV M
light
is available at
http://www-ai.cs.uni-dortmund.de/svm light
- 1
- 2
- 3
前往页