Path: utzoo!attcan!uunet!datran2!smb
From: smb@datran2.uunet (Steven M. Boker)
Newsgroups: comp.ai.neural-nets
Subject: Re: References wanted on Measure and Probablity
Summary: Best one I've found.
Message-ID: <434@datran2.uunet>
Date: 16 Mar 90 13:56:13 GMT
References: <1990Mar7.130350.21423@newcastle.ac.uk> <2945@umbc3.UMBC.EDU> <21740@netnews.upenn.edu>
Distribution: comp.ai.neural-nets
Organization: Data Transforms, Denver, CO
Lines: 39

In article <21740@netnews.upenn.edu>, ferris@eniac.seas.upenn.edu (Richard T. Ferris) writes:
> 
> In article <2945@umbc3.UMBC.EDU> bruce@atria.gsfc.nasa.gov (Bruce Mount) writes:
> >I, too, would like *ANY* suggestions for a beginning book on neural nets.
> >Something like "Neural nets for idiots" or "Neural nets made so simple
> >even YOU can understand it".  
> 
I've spent the last couple of years reading around in the field.  In January
I read a new one called "Neural Computing: Theory and Practice" by
Wasserman.  I believe its Holt Reinhart if my own net is still holding up.

This book is written for someone that wants to understand the basics of
all of the current theories.  Their strengths and weakness and what makes
one net different from another (this sentence no verb).  Wasserman has
gone to a good deal of trouble to make the diagrammatic style of applied
to each network uniform so that the differences in topology are what stands
out.  Hope that made sense, this terminal has the arrow keys mapped to
something that this editor doesnt understand.

Wasserman's insights into each network system are, to my mind, right on
target.  And for those who find matrix notation obscure, he's included the
five minutes up to speed guide to matrix algebra.  All in all, its easily
the best intro I've seen.

Next step would be Rumelhart & McClelland.  Thats the classic of the eighties
as far as I can see.  The Neurocomputing book is a great survey of the
literature going back to William James.  Neurocomputing: Foundations of
Research, Anderson & Rosenfeld.  For the ART perspective you might give a
shot at Stephen Grossberg's Neural Networks and Natural Intelligence.

Good luck.

Steve.

-- 
 #====#====#====#====#====#====#====#====#====#====#====#====#====#====#====#
 #  Steve Boker                #  Black holes are how God divides by zero.  #
 #  smb@datran2.uunet.uu.net   #  ....I have my own methods.                #
 #====#====#====#====#====#====#====#====#====#====#====#====#====#====#====#