Information Technology Reference
In-Depth Information
traditional AI, solves problems through reasoning based on knowledge.
Computational intelligence solves problems based on connections trained from
example data. Artificial Neural Networks, Genetic Algorithms, Fuzzy Systems,
Evolutionary Programming, Artificial Life, etc. are included in computational
intelligence.
Presently, traditional AI mainly focuses on knowledge based problem solving.
In the practical point of view, AI is the science of knowledge engineering: taking
knowledge as the object and investigating knowledge representation, acquisition
and application. This topic mainly introduces and discusses traditional AI. For
computational intelligence, please refer to the topic “Neural Networks” by
Zhongzhi Shi (Shi, 2009).
1.5 Research Approaches of Artificial Intelligence
During the development of AI since the 1950's, many academic schools have
been formed, each holding its specific research methodologies, academic views
and research focuses. This section introduces some research methodologies of AI,
focusing mainly on the cognitive school, logical school, and behavioral school.
1.5.1 Cognitive School
Cognitive school, with representative researchers such as Herbert Simon, Marvin
Minsky and Allen Newell, focuses on functional simulation with computers
based on human noetic activities. In the 1950's, Newell and Simon advocated the
“heuristic program” together, and worked out the “Logic Theorist” computer
program to simulate the thinking process of mathematical theorem proving. Then
in the early 1960's, they developed the “General Problem Solver (GPS)”, which
simulates the common principles of human problem solving with three steps: first,
set the initial problem solving plan; then, apply axioms, theorems and rules to
solve the problems according to the plan; continually proceed with the
means-end analysis, and modify the problem solving until the goal is achieved.
Thus the GPS possesses certain universality.
In 1976, Newell and Simon proposed physical symbol system premise, and
stated that a physical symbol system has the necessary and sufficient means for
general intelligent action. Thus, an information processing system can be viewed
as a concrete physical system, such as human neural system, computer
Search WWH ::




Custom Search