libsvm tools.mht

A SVM classifier, with detail documents.

<h3> Regression Data Sets</h3>=0A=
Please check LIBSVM data shown above.=0A=
=0A=
 Please check the following   <a href=3Dregression_data>directory</a>.=0A=
It includes data used in the following paper=0A=
<p>=0A=
C.-C. Chang and <B>C.-J. Lin</B>.=0A=
<A HREF=3D"../papers/newsvr.ps.gz">=0A=
Training=0A=
nu-support vector regression:=0A=
theory and algorithms=0A=
</A>, =0A=
<I><A =
HREF=3D"http://mitpress.mit.edu/journal-home.tcl?issn=3D08997667">Neural =
Computation</A></I>, 14(2002), 1959-1977.=0A=
=0A=
<p>=0A=
Please check the paper about where these sets are from.=0A=
We really thank those who provided them.=0A=
=0A=
=0A=
<hr>=0A=
=0A=
<h3> Multiclass Data sets</h3>=0A=
Please check LIBSVM data shown above.=0A=
=0A=
 Please check the following   <a href=3Dt>directory</a>.=0A=
It includes data used in the following paper=0A=
<p>=0A=
C.-W. Hsu and C.-J. Lin.=0A=
<A HREF=3D"http://www.csie.ntu.edu.tw/~cjlin/papers/multisvm.ps.gz">=0A=
A comparison of methods =0A=
for multi-class support vector machines=0A=
</A>, =0A=
<I>IEEE Transactions on Neural Networks</A></I>, 13(2002), 415-425.=0A=
=0A=
<p> Most data sets are from the  <a =
href=3Dhttp://www.liacc.up.pt/ML/statlog/>statlog</a>=0A=
or =0A=
<a href=3Dhttp://www.ics.uci.edu/~mlearn/MLRepository.html>UCI =
repository</a>. =0A=
We really thank their efforts!!!=0A=
=0A=
<p> Each attribute of the training data=0A=
(i.e., files end with .scale) =0A=
is scaled to [-1,1].=0A=
Then the testing data is adjusted accordingly.=0A=
Files end with .t are test files.=0A=
In this paper, due to the time limit, we did not conduct cross=0A=
        validation for selecting parameters. Instead, we separate=0A=
training data to "training" and "validation" sets which=0A=
end with .tr and .val.=0A=
=0A=
<p>=0A=
Author: Chih-Wei Hsu=0A=
=0A=
=0A=
<hr>=0A=
<h3> Binary (i.e. two-class) Data sets</h3>=0A=
Please check LIBSVM data shown above.=0A=
=0A=
 Please check the following   <a href=3Dbinary>directory</a>.=0A=
=0A=
<p> Most data sets are from the  <a =
href=3Dhttp://www.liacc.up.pt/ML/statlog/>statlog</a>=0A=
or =0A=
<a href=3Dhttp://www.ics.uci.edu/~mlearn/MLRepository.html>UCI =
repository</a>. =0A=
ijcnn1 is the first problem of IJCNN challenge 2001, provided by Ford=0A=
Scientific Lab. (using the winner's transformation) =0A=
We really thank their efforts!!!=0A=
=0A=
<p> Files end with .t are test files.=0A=
(please check the description for multi-class data sets for =0A=
details)=0A=
=0A=
<p>=0A=
Author: Chih-Chung Chang, Chih-Wei Hsu, and other group members.=0A=
-->
<HR>
Please contact <A =
href=3D"http://www.csie.ntu.edu.tw/~cjlin/index.html">Chih-Jen=20
Lin</A> for any question. </BODY></HTML>

------=_NextPart_000_000C_01C88811.FECF28F0
Content-Type: image/png
Content-Transfer-Encoding: base64
Content-Location: http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/sfd/sfd.png
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------=_NextPart_000_000C_01C88811.FECF28F0
Content-Type: image/png
Content-Transfer-Encoding: base64
Content-Location: http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/roc/heart_scale-roc.png
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------=_NextPart_000_000C_01C88811.FECF28F0--