Path: utzoo!utgpu!utstat!jarvis.csri.toronto.edu!mailrus!cornell!uw-beaver!Teknowledge.COM!polya!shelby!portia!jessica!palle
From: palle@jessica.Stanford.EDU (Palle Henckel)
Newsgroups: comp.ai.neural-nets
Subject: Might NN be the right tool for analysis of 2-dimensional matrices.
Message-ID: <6724@portia.Stanford.EDU>
Date: 17 Nov 89 05:16:54 GMT
Sender: USENET News System <news@Portia.stanford.edu>
Reply-To: palle@jessica.Stanford.EDU (Palle Henckel)
Distribution: usa
Organization: Stanford University
Lines: 34
Keywords: two dimensional matrices

Dear netters,
I'm just about totally ignorant about neural nets.  But I have a problem
and maybe neural nets are the solution, maybe not.  Here is the problem:

I'm interested in a method - or statistic - which can 
measure the concentration vs. randomness of high values 
in a two dimensional matrix.
Example:

High concentration:
     9     9     9
     1     1     1
     1     1     1
returns the highest value

Medium concentration:
     9     1     1
     1     9     1
     1     1     9
returns a medium sized value

Low concentration (random)
     3     5     3
     7     2     2
     1     4     6
returns a low value

The matrices will be bigger than these, though not 
collosal.  Any suggestion or ideas about methods or 
computer progams would be most appreciated.

Thanks plenty.

Palle@jessica.stanford.edu