📄 libsvm -- a library for support vector machines.mht
字号:
From: <Saved by Microsoft Internet Explorer 5>
Subject: LIBSVM -- A Library for Support Vector Machines
Date: Mon, 17 Mar 2008 09:31:52 -0400
MIME-Version: 1.0
Content-Type: multipart/related;
type="text/html";
boundary="----=_NextPart_000_0000_01C88811.BB9F9B00"
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.3198
This is a multi-part message in MIME format.
------=_NextPart_000_0000_01C88811.BB9F9B00
Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Content-Location: http://www.csie.ntu.edu.tw/~cjlin/libsvm/
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD><TITLE>LIBSVM -- A Library for Support Vector =
Machines</TITLE>
<META http-equiv=3DContent-Type content=3D"text/html; =
charset=3Diso-8859-1">
<META=20
content=3D"An integrated and easy-to-use tool for support vector =
classification and regression"=20
name=3Ddescription>
<META content=3D"Support vector machines (SVM), Support vector machine, =
libsvm"=20
name=3Dkeywords>
<META content=3D"MSHTML 6.00.2900.3268" name=3DGENERATOR></HEAD>
<BODY text=3D#000000 vLink=3D#0000ff link=3D#ff0000 bgColor=3D#ffefd5>
<H2>LIBSVM -- A Library for Support Vector Machines</H2>
<H2>Chih-Chung Chang and <A =
href=3D"http://www.csie.ntu.edu.tw/~cjlin">Chih-Jen=20
Lin</A></H2>
<HR width=3D"100%">
<BR><IMG src=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/new.gif"> =
Version 2.85=20
released on November 6, 2007. We now have a script tools/checkdata.py to =
check=20
if your input data is in a correct format. There are also many small=20
improvements. <!--=0A=
Version 2.86 released on April Fools' Day, 2008. =0A=
We now have svm_scale for java. Note that windows=0A=
binary files are renamed to svm-train.exe =0A=
from svmtrain.exe.=0A=
--><BR><IMG=20
src=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/new.gif"> We now have a =
nice page=20
<A =
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets">LIBSVM =
data=20
sets</A> providing problems in LIBSVM format. <BR><IMG=20
src=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/new.gif"> <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/papers/guide/guide.pdf">A =
practical=20
guide to SVM classification</A> is available now! (mainly written for =
beginners)=20
<BR><IMG src=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/new.gif"> <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvmtools">libsvm tools</A> =
available=20
now! <BR>We now have an easy script (easy.py) for users who know NOTHING =
about=20
svm. It makes everything automatic--from data scaling to parameter =
selection.=20
<BR>The parameter selection tool grid.py generates the following contour =
of=20
cross-validation accuracy. To use this tool, you also need to install <A =
href=3D"http://www.python.org/download/">python</A> and <A=20
href=3D"http://www.gnuplot.info/">gnuplot</A>.=20
<P>
<CENTER><IMG=20
src=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/heart_scale.png"></CENTER=
>To see=20
the importance of parameter selection, please see our <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/papers/guide/guide.pdf">guide</=
A> for=20
beginners. <BR><IMG =
src=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/new.gif">=20
Using libsvm, our group is the winner of <A=20
href=3D"http://www.eunite.org/">EUNITE</A> world wide <A=20
href=3D"http://neuron-ai.tuke.sk/competition">competition on electricity =
load=20
prediction </A>, December 2001. The technique used is the support vector =
regression. <BR><IMG =
src=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/new.gif">=20
Using libsvm, our group is the winner of <A=20
href=3D"http://www.ijcnn.net/">IJCNN</A> Challenge (two of the three=20
competieions).=20
<HR>
<H3>Introduction</H3>
<P><B>LIBSVM </B>is an integrated software for support vector =
classification,=20
(C-SVC, <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#nuandone">nu-SVC</A>), =
regression (epsilon-SVR, <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#nuandone">nu-SVR</A>) =
and=20
distribution estimation (<A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#nuandone">one-class =
SVM </A>).=20
It supports multi-class classification. <!-- The basic algorithm is a =
simplification of both <a =
href=3D"http://www.research.microsoft.com/~jplatt/smo.html">SMO</a>=0A=
by <a href=3D"http://www.research.microsoft.com/~jplatt/">Platt </a>and =
SVMLight=0A=
by=0A=
Joachims. It is also a simplification=0A=
of the <a =
href=3D"http://guppy.mpe.nus.edu.sg/~mpessk/smo_mod.shtml">modification=0A=
2</a> of SMO by <a =
href=3D"http://guppy.mpe.nus.edu.sg/~mpessk/">Keerthi</a>=0A=
et al.=0A=
-->
<P>Since version 2.8, it implements an SMO-type algorithm proposed in =
this=20
paper:<BR>R.-E. Fan, P.-H. Chen, and C.-J. Lin. <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/papers/quadworkset.pdf">Working=
set=20
selection using second order information for training SVM</A>. Journal =
of=20
Machine Learning Research 6, 1889-1918, 2005. You can also find a pseudo =
code=20
there. (<A =
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/faq.html#f203">how to=20
cite LIBSVM</A>)=20
<P><FONT color=3D#ff0000>Our goal is to help users from other fields to =
easily use=20
SVM as a tool. </FONT><B>LIBSVM </B>provides a simple interface where =
users can=20
easily link it with their own programs. Main features of <B>LIBSVM</B> =
include=20
<UL>
<LI>Different SVM formulations=20
<LI>Efficient multi-class classification=20
<LI>Cross validation for model selection=20
<LI>Probability estimates=20
<LI>Weighted SVM for unbalanced data=20
<LI>Both C++ and Java sources=20
<LI><A href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#GUI">GUI</A>=20
demonstrating SVM classification and regression=20
<LI><A =
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#python">Python</A>, <A =
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#R">R</A> (also =
Splus), <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#matlab">MATLAB</A>, =
<A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#perl">Perl</A>, <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#ruby">Ruby</A>, <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#weka">Weka</A>, <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#clisp">CLISP</A> and =
<A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#labview">LabVIEW</A> =
interfaces. <A =
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/#csharp">C#=20
.NET</A> code is available. <BR>It's also included in some learning=20
environments: <A href=3D"http://yale.sf.net/">YALE</A> and <A=20
href=3D"http://pcp.sourceforge.net/">PCP</A>.=20
<LI>Automatic model selection which can generate contour of cross =
valiation=20
accuracy. </LI></UL>
<HR>
<H3>Download LIBSVM</H3>The current release (Version 2.85, November =
2007) of=20
<B>LIBSVM </B>can be obtained by downloading the <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/cgi-bin/libsvm.cgi?+http://www.=
csie.ntu.edu.tw/~cjlin/libsvm+zip"><!--=0A=
<a href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/libsvm-2.03.zip">=0A=
-->zip=20
file </A>or <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/cgi-bin/libsvm.cgi?+http://www.=
csie.ntu.edu.tw/~cjlin/libsvm+tar.gz">tar.gz=20
</A>file. <!--(Due to possible slow connection, you may want to download =
it=0A=
from other places:=0A=
<a =
href=3D"http://www.csie.ntu.edu.tw/~cjlin/cgi-bin/libsvm.cgi?+http://www-=
unix.mcs.anl.gov/~lin/libsvm+zip">=0A=
US=0A=
Download</a>). =0A=
-->Please=20
e-mail us if you have problems to download the file.=20
<P>The package includes the source code of the library in C++ and Java, =
and a=20
simple program for scaling training data. A README file with detailed=20
explanation is provided. For <B>MS Windows</B> users, there is a =
subdirectory in=20
the zip file containing binary executable files. Precompiled Java class =
archive=20
is also included.=20
<P>Please read the <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/COPYRIGHT">COPYRIGHT</A>=
notice=20
before using <B>LIBSVM</B>. =20
<HR>
<H3><A name=3DGUI>Graphic Interface</H3>Here is a simple applet =
demonstrating SVM=20
classification and regression.<BR>Click on the drawing area and use =
``Change''=20
to change class of data. Then use ``Run'' to see the results.=20
<CENTER>
<P><APPLET height=3D350 archive=3Dlibsvm.jar width=3D300=20
code=3Dsvm_toy.class></APPLET></CENTER>
<P>Examples of options: -s 0 -c 10 -t 1 -g 1 -r 1 -d 3 <BR>Classify a =
binary=20
data with polynomial kernel (u'v+1)^3 and C =3D 10 <PRE>=20
options:
-s svm_type : set type of SVM (default 0)
0 -- C-SVC
1 -- nu-SVC
2 -- one-class SVM
3 -- epsilon-SVR
4 -- nu-SVR
-t kernel_type : set type of kernel function (default 2)
0 -- linear: u'*v
1 -- polynomial: (gamma*u'*v + coef0)^degree
2 -- radial basis function: exp(-gamma*|u-v|^2)
3 -- sigmoid: tanh(gamma*u'*v + coef0)
-d degree : set degree in kernel function (default 3)
-g gamma : set gamma in kernel function (default 1/k)
-r coef0 : set coef0 in kernel function (default 0)
-c cost : set the parameter C of C-SVC, epsilon-SVR, and nu-SVR (default =
1)
-n nu : set the parameter nu of nu-SVC, one-class SVM, and nu-SVR =
(default 0.5)
-p epsilon : set the epsilon in loss function of epsilon-SVR (default =
0.1)
-m cachesize : set cache memory size in MB (default 100)
-e epsilon : set tolerance of termination criterion (default 0.001)
-h shrinking: whether to use the shrinking heuristics, 0 or 1 (default =
1)
-b probability_estimates: whether to train a SVC or SVR model for =
probability estimates, 0 or 1 (default 0)
-wi weight: set the parameter C of class i to weight*C, for C-SVC =
(default 1)
The k in the -g option means the number of attributes in the input data.
option -v randomly splits the data into n parts and calculates cross
validation accuracy/mean squared error on them.
</PRE>
<P>To install this tool, please read the README file in the package. =
There are=20
Windows, X, and Java versions in the package. <BR>
<HR>
<H3>Additional Information (<A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/faq.html#f203">how to =
cite=20
LIBSVM</A>)</H3><A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/libsvm/faq.html">Frequently =
Asked=20
Questions (FAQ)</A>=20
<P>References of <B>LIBSVM</B>:=20
<UL>
<LI>A guide for beginners: <BR>C.-W. Hsu, C.-C. Chang, <B>C.-J. =
Lin</B>. <A=20
href=3D"http://www.csie.ntu.edu.tw/~cjlin/papers/guide/guide.pdf">A =
practical=20
guide to support vector classification </A>
<LI>For implementation details of <B>LIBSVM</B>, please see the =
document=20
<BR><A =
href=3D"http://www.csie.ntu.edu.tw/~cjlin/papers/libsvm.pdf">LIBSVM : a=20
library for support vector machines. <BR>pdf</A>, <A=20
=
href=3D"http://www.csie.ntu.edu.tw/~cjlin/papers/libsvm.ps.gz">ps.gz</A>.=
<BR>or=20
<BR>R.-E. Fan, P.-H. Chen, and C.-J. Lin. <A=20
=
href=3D"http://www.csie.ntu.edu.tw/~cjlin/papers/quadworkset.pdf">Working=
set=20
selection using the second order information for training SVM</A>. =
Journal of=20
Machine Learning Research 6, 1889-1918, 2005. You can also find a =
pseudo code=20
there. <!--<li>=0A=
Numerical experiments on nu-SVM using <b>LIBSVM</b> can be found in=0A=
<a =
href=3D"http://www.csie.ntu.edu.tw/~cjlin/papers/newsvm.ps.gz">Training=0A=
nu-Support Vector Classifiers: Theory and Algorithms</a>. =0A=
<i>Neural Computation</i> 13(9), 2001, 2119-2147.</li>=0A=
-->
<LI><A=20
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -