www.pudn.com > SVM_SteveGunn.rar > README, change:2001-10-12,size:2642b


Matlab Support Vector Machine Toolbox
-------------------------------------

Author: Steve Gunn

This toolbox was designed as a teaching aid, which matlab is
particularly good for since source code is relatively legible and
simple to modify. However, it is still reasonably fast if used
with the supplied optimiser. However, if you really want to speed
things up you should consider compiling the matrix composition
routine for H into a mex function. Then again if you really want
to speed things up you probably shouldn't be using matlab
anyway... Get hold of a dedicated C program once you understand
the algorithm.

Enjoy!

Version Info
------------

Version: 2.1, 12/10/2001 - interior point QP optimiser added
Version: 2.0, 01/08/1998 - Bug Fixes
Version: 1.0, 10/02/1998 - Initial release

Licence
-------

The Support Vector Machine Toolbox is ONLY available for academic
purposes. It is not available for industrial or commercial
applications of any kind without explicit arrangement with the
author. The software must not be posted on any WWW or ftp sites or
distributed in any other way without prior permission of the
author. The author disclaims all warranties with regard to this
software, including all implied warranties of merchantability and
fitness. In no event shall the authors be liable for any special,
indirect or consequential damages or any damages whatsoever
resulting from loss of use, data or profits, whether in an action
of contract, negligence or other tortious action, arising out of
or in connection with the use or performance of this software.
Permission to sell this software is not granted.

Installation
------------

The distribution now comes in a zip file (partly due to some
problems people were having trying to open tar files with winzip).
Unzip the toolbox under the matlab toolbox directory and add

  ......./matlab/toolbox/svm

to the matlab path. If you are running under a windows OS you
should be ready to go. On an alternative OS you will need to build
the optimiser.

NOTE: The matlab optimisation toolbox also currently contains a qp
program, although it says that it will be replaced by a quadprog
in the future. Make sure that the svm toolbox path comes before at
the front of the matlab path, and it will then use the routine
supplied with the svm toolbox which should be more efficient.

Optimiser
---------

Go into the optimiser directory and type,

  mex -v qp.c pr_loqo.c
  mv qp.mex??? ..

which will build the optimiser for your OS, where the extension
.mex??? will vary depending upon your OS. Move this file up one
directory or add the optimiser directory to the path as well