Temporal Data Mining

82    ◾    Temporal Data Mining

In [Fun02], an incremental SVM classier is proposed that utilizes the

proximity of each class point to one of the two parallel planes. e advan-

tages of the proposed algorithm are as follows:

Ability to add new data very easily•

Ability to compress large datasets with a high compression rate•

Applicability in handling massive data sets eectively•

In [Aio05], a multiclass classier that is able to support a number of

prototypes per class is built using SVM. is is a nonconvex problem and

a greedy optimization algorithm is used that is able to nd local optimal

solutions. e experimental results showed that usi ...