Modern optimization problems decision made using neural network Hopfield; Технологии Microsoft в теории и практике программирования

Λεπτομέρειες βιβλιογραφικής εγγραφής
Parent link:Технологии Microsoft в теории и практике программирования.— 2015.— [С. 87-89]
Κύριος συγγραφέας: Hatem Hassanin
Συγγραφή απο Οργανισμό/Αρχή: Национальный исследовательский Томский политехнический университет (ТПУ) Институт кибернетики (ИК) Кафедра прикладной математики (ПМ)
Άλλοι συγγραφείς: Berestneva O. G. Olga Grigorievna (научный руководитель)
Περίληψη:Заглавие с титульного листа.
The purpose of writing this paper was to study the solution of optimization problems using Hopfield neural network in Matlab environment in order to improve that Neural network as artificial intelligence best method for provide the solutions for optimization problems. This purpose can be achieved through the following steps: 1. Formation of the basic operation of neural networks; 2. Allocation of the problems encountered when solving optimization problems using Hopfield neural network using Matlab [1]; Neural network operates cyclically. Each of the four Hopfield neural network has outputs a signal, which is input, to all other neurons but himself, however, this network cannot be taught almost anything. Network consisting of N neurons cannot remember more than ~ 0.15 * N images. Therefore, the real network should contain enough impressive number of neurons. This is one of the major flaws of the Hopfield network - a small container. In addition to all the images, do not need to be very similar to each other, or in some cases perhaps looping for recognition.
Γλώσσα:Αγγλικά
Έκδοση: 2015
Σειρά:Математическое моделирование и технологии высокопроизводительных вычислений
Θέματα:
Διαθέσιμο Online:http://earchive.tpu.ru/handle/11683/23815
http://www.lib.tpu.ru/fulltext/c/2015/C28/037.pdf
Μορφή: Ηλεκτρονική πηγή Κεφάλαιο βιβλίου
KOHA link:https://koha.lib.tpu.ru/cgi-bin/koha/opac-detail.pl?biblionumber=614547

Παρόμοια τεκμήρια