|
ИСТИНА |
Войти в систему Регистрация |
ИПМех РАН |
||
Comparative Analysis of Methods for Determination of Significance of Input Variables in Neural Network Modeling: Procedure of Comparison and its Application to Model Problemsдоклад на конференции
- Авторы: Гужва А.Г., Шугай Ю.С., Персианцев И.Г., Доленко С.А.
- Международная Конференция : 8th International Conference "Pattern Recognition and Image Analysis: New Information Technologies" (PRIA-8-2007)
- Даты проведения конференции: 2007
- Тип доклада: не указан
- Докладчик: не указан
- Место проведения: Yoshkar-Ola
-
Аннотация доклада:
Procedure of comparative analysis of methods for determination of significance of input variables in neural network solution of regression problem is considered. Variations of this procedure as applied to different methods are described. The suggested procedure is tested on a number of model problems. As the result of testing, the sequence of application of methods for analysis of significance of input variables when considering new problems is suggested.
- Добавил в систему: Доленко Сергей Анатольевич