Information Technology Reference
In-Depth Information
computer systems for noetic activities such as perception, reasoning, learning,
association, decision making, etc., and solves complex problems that only human
experts can solve.
In the history of AI research, different levels of thought are studied from
different views of Symbolicism, Connectionism and Behaviorism.
Symbolicism is also known as traditional AI. It is based on the physical
symbol system hypothesis proposed by Alan Newell and Herbert Simon, which
states that a physical symbol system has the necessary and sufficient means for
general intelligent action. A physical symbol system consists of a set of entities,
called symbols, which are physical patterns that can occur as components of
another type of entity called an expression (or symbol structure). The system also
contains a collection of processes that operate on expressions to produce other
expressions: processes of creation, modification, reproduction and destruction.
Connectionism, also known as neural computing, focuses on the essentials
and capabilities for non-programmatical, adaptative and brain-like information
processing. The research field is rapidly developing in recent years, with a great
number of neural network mechanisms, models and algorithms emerged
continuously. Neural network systems are open neural network environments
providing typical and practically valuable neural network models. The open
system enables convenient adding of new network models to the existing system,
so that new network algorithms can be debugged and modified with the friendly
user interfaces and varieties of tools provided by the system. Moreover, it is also
convenient to improve existing network models, thus the system provides
excellent environment to develop new algorithms.
Neural computing investigates the brain functionalities based on the nervous
system of human brains, and studies the dynamic actions and collaborative
information processing capabilities of large numbers of simple neurons. The
research focuses on the simulation and imitation of human cognition, including
processes of perception and consciousness, imagery thought, distributed memory,
self-learning and self-organization. Neural computing is particularly competent
in parallel search, associative memory, self-organization of spatio-temporal data
statistical descriptions, and automatic knowledge acquisition through interrelated
activities. It is generally considered that neural networks better fitted low level
pattern processing.
Basic characteristics of neural networks include: a. distributed information
storage, b. parallel information processing, c. capabilities of self-organization and
self-learning (Shi, 1993). Owing to these characteristics, neural networks provide
