Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uflorida!usfvax2!pollock
From: pollock@usfvax2.EDU (Wayne Pollock)
Newsgroups: comp.ai.neural-nets
Subject: Re: implementation of "traveling salesman algorithm" on connection machine
Keywords: TSP, connection machine
Message-ID: <1236@usfvax2.EDU>
Date: 8 Feb 89 17:22:23 GMT
References: <1637@cps3xx.UUCP> <1233@usfvax2.EDU> <6173@phoenix.Princeton.EDU> <18090@glacier.STANFORD.EDU>
Reply-To: pollock@usfvax2.UUCP (Wayne Pollock)
Distribution: usa
Organization: University of South Florida at Tampa
Lines: 23

In article <18090@glacier.STANFORD.EDU> jbn@glacier.STANFORD.EDU (John B. Nagle) writes:
>
>       The same hype appeared in an article in Business Week.  Comparing
>the time required to find a near-optimum solution with a hardware neural
>net to that required to find an optimal solution on a Cray is misleading,
>if not fraudulent.  ...
> [description of simple algorithm which does the same job quickly on
>  traditional hardware]
>
>       [therefore] You don't need a Connection Machine for this.
>
>					John Nagle

I think you're missing the point.  You're quite right, you don't need
a connection (neural network) machine for this.  The point is, it can
be done on a connection machine.  It is interesting to see what other
problems can be solved on a connection machine, that have generally
thought to be hard or difficult.  Perhaps some insight to these problems
or to computing in general might result.

Wayne Pollock (The MAD Scientist)	pollock@usfvax2.usf.edu
Usenet:		...!{uflorida, codas}!usfvax2!pollock
GEnie:		W.POLLOCK