|
Article information
2003 , Volume 8, ¹ 6, p.31-37
Ignatjev N.A., Madrakhimov S.
Some methods for neural networks transparency increase
The measure of a structural variety of the neural network representing noncrossing classes of objects is introduced. A method of selection of the sets of information with various types of attributes, based on a calculation of the contribution of each attribute to the division of objects of different classes, is described.
[full text] Classificator Msc2000:- *92B20 Neural networks, artificial life and related topics
Keywords: neuron, learning sample
Author(s):
Ignatjev NikolayA.
Dr. , Associate Professor
Position: Professor
Office: National University of Uzbekistan
Address: 100022, Uzbekistan, Tashkent
Phone Office: (99871) 246-16-27
E-mail: n_ignatev@rambler.ru
Madrakhimov Shavkat
PhD. , Associate Professor
Position: Head of Chair
Office: National University of Uzbekistan
Address: 100093, Uzbekistan, Tashkent
E-mail: mshavkat@yandex.ru
Bibliography link:
Ignatjev N.A., Madrakhimov S. Some methods for neural networks transparency increase // Computational technologies. 2003. V. 8. ¹ 6. P. 31-37
|
|
|
|
