[Visinfo] Review of JOONE an alternative Kohonen SOM software pkg

[Mike Godwin]

After the SOM_PAK tutorial, I was curious to see if there were any more
recent and possibly handier programs for implementing the kohonen
algorithm.  Angus pointed me towards Joone (Java Object Oriented Neural
Engine) so I downloaded it and went through the tutorials...

http://www.jooneworld.com/

On the pros: it's java based and therefore easily run from a variety of
platforms.  It has a very clear and intuitive graphic interface -- one
of those drag and drop style programming interfaces, ie. draw a box that
represents your input, a box that represents the output, and some boxes
that represent your neural network, connect all the arrows and voila! 
It's really well-suited to visualizing complex neural networks and
assessing their learning strengths and faults.

Unfortunately Joone is definitely not set up for creative/alternative
visualization of the output.  It is relatively easy to use the editor to
read a table, utilize come customized kohonen algorithm, and then output
a table; but it would still be up to you to take that table and parse it
into something attractive if data visualization were the ultimate objective.

In the end I think this is a wonderful tool for someone more interested
in artificial intelligence and neural networks, and aside from browsing
the core engine for the implementation of the kohonen algorithm I'll be
passing it up for my own work in this class.