From: anon03e2@nyx.cs.du.edu (Fallen Angel)
Date: Tue, 18 May 93 14:41:46 PDT
To: cypherpunks@toad.com
Subject: Neural Nets to decrypt?
Message-ID: <9305181912.AA05130@nyx.cs.du.edu>
MIME-Version: 1.0
Content-Type: text/plain


I am naive on a couple of fronts, so apologies if this is off-base.

As computing evolves, would it be very possible to use a neural net
or like system to design an ideal code-breaking engine? From my u
understanding, code-breakers are only as good as their designer's
knowledge of the algorithim. Such a hypothetical system would develop
its own cracking alogorithim which could be more efficient by several
factors.
  
I think another variation would the use of GA's to create a more efficient
code-breaker. I know that GAs have been used to achieve near-ideal results
in least-step sorting algorithims. Would code-breaking be the same
in principle?