Getting Started with VFML

At its core, VFML contains a library that you will need to compile and link with your programs.  VFML has a set of interfaces through which you access the functionality of the library.  It also contains a few programs which should help you with your machine learning tasks.

The best way to get the most recent version of VFML is to download it from our project page. You can also get the most recent code from the CVS server there. For more information see The module name is vfml so the following commands should work (simply press 'enter' if asked for a password):

cvs login

cvs -z3 co -P vfml

Getting Started with UNIX

VFML was developed and tested with Linux. It should compile easily under any UNIX. If it doesn't work for you please send us a bug report. Once you have the source take the following steps to compile it:

  1. change directory into the distribution directory
  2. edit the makefile so that the correct arch is uncommented (try Unix if none of the others fit)
  3. type make
  4. add the <distribution>/bin directory to your path

Once that is done, you might like to check out the creating-learners page, which contains a starter project and makefile. It should help get you started.

Getting Started with Windows

If you have Cygnus installed, follow the UNIX getting started instructions (if you don't have Cygnus installed you might consider it - it's free!)

VFML was developed on Linux, but is written with standard C and should work in other environments. For non-cygnus windows try the following:

The implement-learner example contains a VC++6.0 project which should help get you started.  You might need to change some of the configurations and paths depending on where you installed the VFML distribution - the sample project file assumes you installed VFML as c:/proj/vfml.   Here are the steps that were used to create this VC++ project:  

Getting Started with C++

VFML will work with C++, just remember that you need to let your compiler know that VFML is written in C.  For example, you need to use the following statement to include vfml.h into a C++ files:

extern "C" {
   #include "vfml.h"

Getting Datasets

We have converted a bunch of the UCI-repository uci datasets to C4.5 format for use with VFML.  You should download them, and put them in the same directory that contains the VFML distribution directory.