Frank Rosenblatt (11
July 1928– 1971) was a New York City born computer scientist who
completed the Perceptron, or MARK 1, computer at Cornell University in
1960. This was the first computer that could learn new skills by trial
and error, using a type of neural network that simulates human thought
processes.
Rosenblatt’s perceptrons were initially simulated on an IBM 704
computer at Cornell Aeronautical Laboratory in 1957. By the study of
neural networks such as the Perceptron, Rosenblatt hoped that "the
fundamental laws of organization which are common to all information
handling systems, machines and men included, may eventually be
understood."
A 1946 graduate of the Bronx High School of Science, Rosenblatt was a
colorful character at Cornell in the early 1960s. A handsome bachelor,
he drove a classic MGA sports car and was often seen with his cat
named Tobermory. He enjoyed mixing with undergraduates, and for
several years taught an interdisciplinary undergraduate honors course
entitled "Theory of Brain Mechanisms" that drew students equally from
Cornell's Engineering and Liberal Arts colleges.
This course was a mélange of ideas drawn from a huge variety of
sources: results from experimental brain surgery on epileptic patients
while conscious, experiments on measuring the activity of individual
neurons in the visual cortex of cats, studies of loss of particular
kinds of mental function as a result of trauma to specific areas of
the brain, and various analog and digital electronic circuits that
modeled various details of neuronal behavior (i.e. the perception
itself, as a machine).
There were also some breathtaking speculations, based on what was
known about brain behavior at this time (well before the CAT or PET
scan was available), including one calculation that, based on the
number of neuronal connections in a human brain, the human cortex had
enough storage space to hold a complete "photographic" record of its
perceptual inputs, stored at the 16 frames-per-second rate of flicker
fusion, for about two hundred years.
In 1962 Rosenblatt published much of the content of this honors course
in the book "Principles of neurodynamics: Perceptrons and the theory
of brain mechanisms" (Spartan Books, 1962) which he used thereafter as
a textbook for the course.
Research on similar devices was also being done in other places such
as SRI, and many researchers had big expectations on what they could
do. The initial excitement became somewhat reduced, tough, when in
1969 Marvin Minsky and Seymour Papert published the book "Perceptrons"
with mathematical proofs that elucidated some of the characteristics
of the three-layer feed-forward perceptrons. For one side, they
demonstrated some of the advantages of using them on certain cases.
But they also presented some limitations. The most important one was
the impossibility of implementing general functions using only "local"
neurons, that don't have all inputs available. This was taken by many
people as one of the most important characteristics of perceptrons.
Rosenblatt died in a boating accident in 1971. After research on
neural networks returned to the mainstream in the 1980s, new
researchers started to study his work again. This new wave of study on
neural networks is interpreted by some researchers as being a
contradiction of hypotheses presented in the book Perceptrons, and a
confirmation of Rosenblatt's expectations, but the extent of this is
questioned by some. [1]
In 2004 the IEEE established the Frank Rosenblatt Award, for
"outstanding contributions to the advancement of the design, practice,
techniques or theory in biologically and linguistically motivated
computational paradigms including but not limited to neural networks,
connectionist systems, evolutionary computation, fuzzy systems, and
hybrid intelligent systems in which these paradigms are contained."
[2]
See also
Artificial neural networks
History of artificial intelligence
AI Winter
Perceptrons
Mark-I as Mark-II were not simulated with the help of computer. It was
electo-mechanical device. See Rosenblatt reports. Computer-based
perceptrons were simulated by M. Minsky and other researchers. |
|
|