Article: 2 of 6 

ARTNO       0021200-0

ARTICLE        automata, theory of

TEXT   {awt-oh-mat'-uh\}

The theory of automata deals with
the fundamental behavioral principles
of automatic machines.  It provides
the theoretical framework for automated
processes in which sensors and detectors
replace human sense organs, actuators
powered by electric motors or hydraulic
forces replace human muscles, and
microprocessors replace the human
brain.  Automata theory was once
largely an abstract branch of mathematics.
It was an attempt to define the behavior
of machines or systems in terms 
    of data inputs and outputs. Specific
areas of interest include    computing
machine theory, language or grammar
theory, logic theory, and ARTIFICIAL
INTELLIGENCE .  Recently, automata
theory has been implemented in practical
computer-based operations that mimic
human processes. 

In 1936, Alan TURING, an English
mathematician, systematized   automata
theory by developing, on paper, a
model for the digital computer. Turing
proved that his "universal computer"
could solve virtually any mathematical
or logical problem that could be
formulated in a logically consistent
manner.  He also showed that certain
types of mathematical problems were
beyond its powers.

In the 1940s, Warren McCulloch and
Walter Pitts, of theMassachusetts
Institute of Technology (MIT), developed
the theory of neural net automation.
Essentially an abstract model  of
the human nervous system, neural
net automation theory describes a
system of neurons, that, when stimulated
above acertain threshold, will fire,
and transmit an impulse along connecting
fibers.  Neural net automation theory
has been used

in a variety of automata applications.

The 1950s brought a surge of enthusiasm
for automata theory.  American linguist
Noam CHOMSKY showed that automated
language analysis could be performed
on both human language and programming
languages.  John McCarthy of MIT
coined the term artificial intelligence
and defined it as "the science of
making computers do things which
if done by men would require intelligence."
This clever definition sidestepped
a troubling issue:  whether or not
computers can actually think.  Claude
Elwood SHANNON, also of MIT, outlined
methods whereby a machine could be
programmed to play winning chess
and learn from its      experiences.
Additional advances were made by
the researcher Herbert A.  SIMON,
who developed the "General Problem
Solver," a program that was able
to prove a well-known theorem in
a way that was simpler than traditional
methods, and by Terry Winograd, who
developed a program that could understand
both verbal language and complex
geometric manipulations.

Automata theorists have struggled
with the difficulties of learning
how machines might be made to see,
to hear, to learn from experience,
and to think.  Critics have accused
researchers in the field of producing
few results, and some have predicted
that it will be impossible for machines
to mimic real human thinking.  Nevertheless,
automatic machines that perform routine
tasks have become commonplace in
industry.

Active areas in modern automata theory
include studies in expert systems
and man-machine interactions (see
CYBERNETICS). Expert systems are
computer programs that to some extent
encapsulate the expertise of highly
trained professionals Caduceus, for
example, is a knowledge-based system
that diagnoses blood diseases from
1,000 programmed rules, and Prospector
is used to locate mineral deposits
using geological\        data supplied
by its programmers.

One possibly rich field for basic
research is cellular automata.  First
elucidated by mathematicians John
VON NEUMANN and Stanislaw Ulam in
the early 1950s, cellular automata
are made up of mathematical "cells"--squares
on a chessboard offer a good analogy--that
change their value, or state, according
to simple rules, or ALGORITHMS. The
state of each "cell" is affected
by the states of neighboring cells.
Researchers in       this field have
used computers to model such natural
processes   as snowflake formation.
Some automata theorists believe it
possible to simulate complex physical
systems, including the human nervous
system , through cellular automata.

Advances in applied areas, such as
VOICE RECOGNITION and      machine-learning,
have been frustrated by the slow
speeds of     present-day computers.
Higher speeds are needed in order
to  handle the data involved in such
processes.  As an example, the creation
of a fully autonomous land vehicle
that can avoid obstacles and move
safely at freeway speeds will require
computer processing speeds 100 to
10,000 times greater than the  maximum
speeds now possible. PARALLEL PROCESSING
techniques that mimic the processes
of the human brain, and newer  generations
of smaller, faster microelectronics
will be needed      to attain such
goals.  DONNA AND TOM LOGSDON

Bibliography:  Arbib, Michael, Computers
and the Cybernetic\        Society,
2d ed.  (1984);  Logsdon, Tom, Computers
Today and\        Tomorrow (1985);
McCorduck, Pamela, Machines Who Think
(1981).\ \ Copyright notice:

Copyright by Grolier Electronic Publishing,
Inc.

                  ---- End of Article
----