Neural network stacks?

[Marielle Lange]

> I'm starting to learn about neural networks and hoping some stacks  
> are out
> there somewhere. Any leads are appreciated!

No stack... this is too specialized a topic.

<http://revolution.lexicall.org/wiki/tiki-index.php? 
page=SoftwareModelisation>

If you are complete beginner, best way to learn is to download tLearn.
<http://crl.ucsd.edu/innate/tlearn.html>

It's quite intuitive to use. You define a network with a config file  
like this:
NODES:
nodes = 161
inputs = 105
outputs = 61
output nodes are 101-161
CONNECTIONS:
groups = 0
1-100     from     i1-i105
101-161        from    1-100
1-161 from 0
SPECIAL:
selected = 1-100
weight_limit = 0.2

There is an associated book: Plunkett, K., & Elman, J. L. (1997).  
Exercises in rethinking innateness. Cambridge, MA: MIT Press. which  
takes you step by step, from complete beginner to intermediate. It is  
found in most of the academic bookstores.... and hum, if you use  
google you can find it for free on the web:
http://crl.ucsd.edu/~elman/Courses/cog202/tlearn.html
http://cspeech.ucd.ie/~connectionism/
http://www.cogsci.ucsd.edu/~rik/courses/readings/plunkett97-RIEg/


If are serious about "understanding" the contribution of neural  
networks, the two "bibles" are (and remain):

McClelland, J. L., Rumelhart, D. E., & Group, t. P. R. (1986).  
Parallel Distributed Processing. Volume 2. Cambridge, MA: MIT Press..

Rumelhart, D. E., McClelland, J. L., and the PDP Research Group.  
(1986). Parallel Distributed Processing: Explorations in the  
microstructure of cognition: Volume 1. Foundations. Cambridge, MA:  
MIT Press..

There was an accompanying exercise book:
McClelland, J. L., Rumelhart, D. E., & Hinton, G. E. (1986). The  
appeal of parallel distributed processing. In D. E. Rumelhart & J. L.  
McClelland (Eds.), Parallel distributed processing: Explorations in  
the microstructure of cognition. Cambridge, MA: MIT Press..

Which I strongly recommend.

The book advertised at:
http://www.cnbc.cmu.edu/Resources/PDP++//PDP++.html
Is also excellent, but this is for a quite advanced public. The  
software presented there is for a *very* advanced public (you should  
understand the basics of neural networks before using it).

If you need something more specific, contact me privately (I have  
been teaching connectionism in the past, from a cognitive psychology  
point of view... I even have taken Jay McClelland for a tour of  
Brussels).  I have a few lecture slides and handouts with exercises  
and solutions in French and English as well.

Ah yes, Why do you want to learn about neural networks. They were hot  
in the 80s, but nowadays most researchers recognize that some  
"structure" facilitates the learning.

Marielle
------------------------------------------------------------------------ 
-------------------
Marielle Lange (PhD),  Psycholinguist

Alternative emails: mlange at blueyonder.co.uk, M.Lange at ed.ac.uk
Homepage:  http://homepages.lexicall.org/mlange/
Lexicall: http://lexicall.org
Revolution-education: http://revolution.lexicall.org