
42 STATISTICAL LEARNING AND DATA SCIENCE
1
2
3
4
5
6
7
8
9
10
20001000500300100
mean error
training size
TAG1
SVM
SVM+
SVM*
7.5
8
8.5
9
9.5
10
10.5
11
11.5
12
12.5
20001000500300100
mean error
training size
TAG2
SVM
SVM+
SVM*
4
4.5
5
5.5
6
6.5
7
7.5
20001000500300100
mean error
training size
TAG3
SVM
SVM+
SVM*
FIGURE 3.4
Learning curves of SVM and SVM+ in music classification.
3.6 Conclusions
We presented a framework of sparse co njugate direction optimization algorithms. This
framework combines the ideas of the SMO algorithm a nd optimization using a conjugate
direction. Our framework is very g e neral and can be efficiently instantiated to any optimiza-
tion problem with ...