Dear colleagues,
I have sent this letter to several of colleagues like you, for whom I
know that there is an interest in training SVMs.
This is one more routine, on the list of few ones, for training linear
SVM on ultra huge data sets which in addition may be of high
dimensionality.
LinearSVM is able to handle both binary and multiclass problems easily.
It's accurate, it's fast, it's user friendly, and it can be downloaded
from here:
http://www.linearsvm.com
or from here
http://www.people.vcu.edu/~vkecman/Downloads.html
LinearSVM shows very good performance for dense data sets.
Sorry if you feel bothered, but I believe that you and/or your younger
co-workers may be interested in testing it.
Best regards,
Vojo
--
*...........*...........*...........*...........*
Vojislav Kecman
VCU Engineering, Computer Science
401 West Main, # E4244
P.O.Box 843019
Richmond, VA 23284-3019, USA
Phone: +1 804 827 3608
Fax: +1 804 828 2271
. . . . . . . . . . . . . . . . . . . . My websites are below:
http://www.people.vcu.edu/~vkecman VCU LAAL site
http://www.support-vector.ws book 1
http://www.learning-from-data.com book 2
|