Industrial Artificial Intelligence
Rosenblatt’s Perceptron 6 had three layers
with the middle set of neurons called the
association layer. In 1969 Marvin Minsky and
Seymour Papert from MIT wrote a book 7
stating that a multi-layered approach to
neural networks would not be successful,
which effectively eliminated funding for that
kind of AI. Tell that to the folks creating
amazing results with Convolution Neural
Network (CNN) image recognition. It took
twenty years for the innovation to get
started again.
E ARLY I NDUSTRIAL AI
Applications of AI in industry extended the
ideas for pruned searches of the solution
space, especially with the focus on product
design, production planning and logistics.
Mark Fox from Carnegie Mellon University
summarized 8 the advances within two AI
categories: Knowledge Representation and
Search. He framed the driving forces as
shown in Table 1.
Force Description
Expertise Scarcity is endemic to many corporations
Decision complexity Large number of alternatives from which to choose,
exacerbated by flexible manufacturing systems
Decision support Information is becoming more complex, reduce to only
what is necessary for an individual to decide
Decision timelines
Coordination Deadline to plan and decide is decreasing
How can designs help optimize the down-stream
activities?
Table 1: Issues in Manufacturing
These
motivations
drove
specific
applications of AI, especially expert systems,
in the 1970s and 80s. Table 2 8 below
summarizes those early applications.
I NDUSTRIAL AI T ODAY
The ground-breaking innovations in the last
century are having a dramatic impact on
many aspects of our lives. In addition, there
is benefit from revisiting those solutions
6
Rosenblatt, F., “The Perceptron: A Probabilistic Model for Information Storage and Organization in the Brain”, Cornell
Aeronautical Laboratory, Psychological Review, v65 (1958).
7 Minsky, M., Papert, S., “Perceptrons: an introduction to computational geometry”, The MIT Press, Second Edition (1972).
8 Fox, M.S., “Industrial Applications of Artificial Intelligence”, Robotics, Vol. 2, Issue 4, Elsevier (December 1986).
- 3 -
June 2019