2
The Connectionist Explanation of the Mass Mind's Dreams
The secret behind the problem-solving abilities of world views is
the same as that behind the success of superorganisms.  It lies in the
power of networks.
One of the most irritating mathematical dilemmas facing
computer scientists is the "traveling salesman problem."  Imagine that
you're a salesman about to go out on a trip.  You've decided to visit ten
different cities.  How would you figure out which town should come
first on your route, which should come second, which should come
next, etc. to give you the shortest mileage?  Simple, you'd just sit down
with a map and have an answer in an instant.  But it's not that easy.
Turns out that the potential number of sequences in which you could
hit these ten destinations comes to 181,440.
Well, how about flicking on the switch of your desktop pc?
Unfortunately, a conventional computer is rather slow at solving
traveling salesman problems.  Normal computers measure each
potential routing one at a time.  With close to 200,000 different options
to test, that's a time-consuming process.  And the traveling salesman
dilemma is one that industry encounters all the time.  Phone
companies laying cable, for example, run into variations on the
conundrum that make the simple ten-city example look like child's
play.  What's to be done?
The stumbling block comes, in part, from the way normal
computers are set up.  Looking at a problem only one small part at a
time, they can't see the big picture.  In the mid-80's, however, a new
breed of computer experts began to experiment with machines that can
"explore" a broader canvas.  These are called "connectionist webs" or
"neural nets."
<<  <  GO  >  >>