Prediction of properties of molecular graphs based on RBF neural networks

Kumskov, M.I.; Vasilyeva, V.O.; Rybakova, E.O.

Авторы: Kumskov Mikhail I., Vasilyeva Varvara O., Rybakova Ekaterina O.
Журнал: IEEE Transactions on Neural Networks and Learning Systems
Том: 24
Год издания: 2023
Аннотация: A method for forming the architecture of an RBF neural network for solving the problem of predicting the properties of molecular graphs is proposed and investigated. The method is based on the identification of the cluster structure of the training dataset in the "structure-property" problem. The results of the method study for predicting the properties of molecules for various variants of the attribute description of molecular graphs are presented. The RBF neurons of the network cover the identified clusters, the placement of neurons is based on the k-means algorithm for each identified cluster.A technique for forming the architecture of a neural network is substantiated based on a preliminary analysis of the training data set and the assignment of RBF neurons to the elements of the training set. The obtained results are comparable with the previously implemented methods, which were based on an evolutionary algorithm - Group Method of Data Handling (GMDH).The paper was published with the financial support of the Ministry of Education and Science of the Russian Federation as part of the program of the Moscow Center for Fundamental and Applied Mathematics under the agreement075-15-2022-284, this research has been supported bythe Interdisciplinary Scientific and Educational School of Moscow University ”Brain, Cognitive Systems, Artificial Intelligence”
Добавил в систему: Кумсков Михаил Иванович

	ИСТИНА	Войти в систему Регистрация
	ИСТИНА ЦЭМИ РАН
	Главная Поиск Статистика О проекте Помощь

ИСТИНА

ИСТИНА ЦЭМИ РАН

Prediction of properties of molecular graphs based on RBF neural networksстатья

Прикрепленные файлы