README
=======================================================================
Incremental and decremental support vector machine learning
Matlab code, data and demos
G. Cauwenberghs
gert@ucsd.edu
=======================================================================

This directory contains Matlab code, data files, and example demos for
incremental SVM classification, including exact leave-one-out (LOO)
cross-validation.  The file incremental.tar.gz contains the entire
gzipped, tarred directory.

To start, run test_2d or test_diabetes at the Matlab prompt.  Make
sure to have all *.m and *.mat files in your directory.  Use the
Matlab "help" function to find syntax and more information on the
implemented functions.  Some of the parameters are available as global
variables in the workspace (see test*.m for examples).

This software is in the public domain; there are no implied warranties
of any kind!  Send bug reports to gert@ucsd.edu

Matlab functions and scripts:
-----------------------------
svcm_*.m     support vector classification machine
kernel.m     kernel function used in svcm_*.m
test*.m      example demo scripts
Set*.m       graphics formatting
gen*.m       data generating functions

Matlab data:
------------
*.mat        vectors x [L,N], and labels y [L,1] (-1 or 1)

Sample output:
--------------
*.eps        encapsulated PostScript

Reference:
----------
G. Cauwenberghs and T. Poggio, "Incremental and Decremental Support
Vector Machine Learning," in Adv. Neural Information Processing
Systems (NIPS*2000), Cambridge MA: MIT Press, vol. 13, 2001.
(http://www.biology.ucsd.edu/~gert/papers/nips00_inc.pdf)

Other incremental SVM code:
---------------------------
SVM incremental on-line classification, LOO evaluation, and
hyperparameter optimization (Chris Diehl):
http://www.cpdiehl.org/incrementalSVM.html

SVM incremental on-line regression (Francesco Parrella):
http://onlinesvr.altervista.org/