Information Technology Reference
In-Depth Information
important for building computational tools that optimally solve problems of high
computational complexity.
Apparently, one of the most promising opportunities to find constructive prin-
ciples for such devices is constituted by heuristic approaches. In particular, it can be
assumed that the ability to effectively carry out operations of high computational
complexity is inherent to systems with highly complex behavior. Therefore, the
problem of designing effective devices for solving problems of high computational
complexity becomes a problem of finding the design principles of systems showing
high behavioral complexity. Given the main factors determining the functioning of
distributed systems, the following principles can be chosen:
Distributed nature of the system, resulting in high parallelism of processes in the
system
The complexity of system's dynamics, which leads to logically complex local
behavior
Layered system architecture
In the following, an attempt will be made to show that consistent use of these
principles in designing information processing devices leads to a new paradigm
which is fundamentally different from the von Neumann paradigm and to optimal
solutions for problems of high computational complexity.
4.2 Computer Engineering and the Problem of Artificial
Intelligence
Modern digital computers, the principles of information processing, and the infor-
mation technology have radically changed the world around us. As a consequence,
in the second half of the last century the dominant paradigm of von Neumann for
constructing information processing devices seemed most optimal, if not the only
possible one. Explicitly or implicitly, it was assumed that the progress of semicon-
ductor technology, combined with new physical ideas, would lead to a further
increase in productivity of digital (in fact, von Neumann) devices that will be
able to solve all or almost all pressing information tasks. Nevertheless, the ever-
growing significance of information for large dynamic systems has substantially
undermined this widely accepted view.
In the 1980s of the last century, the famous American cybernetician Michael
Arbib expressed his belief that further development of computing concepts will
take the path of imitating the style of information processing by the human brain.
“The human brain is a metaphor of the next (sixth)-generation computers”—he
wrote—“Its style is determined by the interaction between systems, many of which
correspond to the joint operation of the spatio-temporal structures in layered neuron
structures.”
