Information Technology Reference
In-Depth Information
autonomous devices capable of replacing the man in critical conditions, to reliably
predict weather, and to regulate the economy and social problems.
In general investigation of the mechanisms of functioning of large dynamic
systems and their external manifestations involves a series of information tasks,
such as:
￿ Recognition of images, scenes, and situations which may be somewhat arbi-
trarily reduced to:
- Classification of objects based on a given set of features
- Object segmentation, i.e., subdividing it into a set of simpler fragments
- Context-based recognition of fragments with subsequent construction of the
source object (important in such areas as medicine, material science, etc.)
￿ Study of evolution of systems with complex dynamics of behaviors (e.g.,
predator-prey problems, evolution of populations of biological cells, etc.)
￿ Selection of the optimal (in some predefined sense) structure or behavior of a
complex multifactor system with a complex decision tree (“the traveling sales-
man problem,” strategic and tactical decision games)
￿ Control problems, which include:
- Continuous recognition of situations
- A continuous selection of the optimal strategy (navigation of autonomous
robots operating in a complex changing environment and other similar
devices)
A fundamental feature of large dynamic systems is their distributed nature. This
means that the system represents a huge collection of simple (for this system)
elements which interact with each other. At the level of the human society such an
element is a single person in the crowd, in physicochemical reaction media—
individual molecular components of the reactions taking place in the system. But
regardless of the level of complexity of behavior of a separate element, distributed
systems are characterized by common structural and behavioral characteristics.
Processes in a distributed system (environment) occur simultaneously at its each
point. In other words, it is characterized by high (or rather gigantic, compared, e.g.,
with modern parallel digital computers) parallelism of the system's actions.
In a large number of distributed systems nonlinear interactions between indi-
vidual elements of the system occur. This leads to much more complex behavior of
the system than that of its individual elements. A simple but impressive example,
frequently used in recent years, is the activity of ant communities. One of the main
functions of the community is the search for and delivery of food to the anthill.
Elementary actions of an individual person are simple: random search, exudation of
specific chemicals (pheromones) once food has been found and is being transported
to the anthill, and following the pheromone trail. These simple mechanisms help
memorize the way to accidentally found food, involve other individuals in this
process, optimize delivery routes, etc. A characteristic feature of distributed sys-
tems with nonlinear interactions of the elements is that the behavior of the system as
a whole is much more complex than that of its individual elements. Moreover, it
Search WWH ::




Custom Search