On consistency of quadratic and bilinear systems

The earlier described set of estimate calculation algorithms (ECA) for solving recognition and classification problems is considered, in particular ECA subclass described by weights of indices determining recognized objects and weights of precedent objects, for which sufficiently full information is known. Solving the related problem of constructing an optimal on a control sample algorithm is reduced to isolation of the maximal consistent subsystem of the system of bilinear inequalities. Construction of relatively simple criteria of consistency of quadratic and bilinear systems for effective solving of the latter problem is reported.

Авторы
Журнал
Номер выпуска
3
Язык
Русский
Страницы
295-297
Статус
Опубликовано
Том
402
Год
2005
Организации
  • 1 Rossijskij Univ. Druzhby Narodov, Moscow, Russian Federation
Ключевые слова
Classification (of information); Efficiency; Optimization; Pattern recognition; Quadratic programming; Bilinear systems; Consistency; Algorithms
Цитировать
Поделиться

Другие записи