| Биогидравлическая
модель нервных систем (21.01.04) |
| Антонов Владимир
Михайлович, ктн, доц. ЛГТУ |
| Излагается модель
нервных систем, в которой принято, что управление
осуществляется не электрическим биотоком, а тем
потоком веществ, который одновременно является
питательным. Рассматривается механизм обучения
систем. |
|
| Представление
структур данных в ассоциативной нейронной среде |
| В. В. Борисов |
| The Associative Neuron Medium are considered.
The analysis of structures of data in Associative Neuron Medium is realized. The
efficiency of using of Associative Neuron Medium in solution of tasks of data presentation
and processing is shown. |
|
| Интегрированная
среда для исследования генетических алгоритмов
DSK-GEN |
| М.С.Куприянов,
Н.И.Матвиенко |
| Genetic algorithms base on the natural
evolution principle and make directed search in a given space. As a result of inefficient
use of some genetic operators (mutation and crossover) destruction of evolutionary process
can be brought in. Special software DSK-GEN was realized for the research purpose of the
various effective structures of genetic algorithms. The description of its modules and
test results is included. Direction of its further development is determined. |
|
| Генетический
подход к оптимизации входного сигнала
динамической системы |
| Л.А Мироновский,
К.Ю. Петрова |
| This paper proposes genetic and combined
algorithms for evaluation of operator norms of dynamic systems and the corresponding
optimal input signals. These algorithms are compared with each other, with von-Mizes
iterative procedure and with method of steepest descent. |
|
| К расчету
топологической пластичности ядерных нейронных
сетей |
| А.Ю. Дорогов |
| In the article the quantitative evaluation of
topological plasticity neural networks is considered. The exact formulas and asymptotic
evaluations are obtained. The example of calculation of topological plasticity degree is
indicated. |
|
| Построение
математических моделей непрерывных
технологических процессов на базе
Neuro-Fuzzy-технологии |
| Ю.А. Кораблев, К.С.
Шаров, М.Ю. Шестопалов |
| In the report we consider a task
of mathematical models synthesis of continuous TP, based on classical and neuro-fuzzy
methods, which combination allows to expand a set of solvable tasks. In particular,
processing and identification of TP are based on the statistical and correlation analysis,
which results are the initial data to make a fuzzy decision about topology of the system.
The turned out topology of the system is simulated with neural networks corresponding to
concrete links. Result is the model of the system constructed on the basis of structured
neural network. |
|
|
|
|
Введение в
теорию нейронных сетей и программная реализация
их основных конфигураций |
|
|
