Volume 6, Issue 1 (Journal of Control, V.6, N.1 Spring 2012)                   JoC 2012, 6(1): 41-50 | Back to browse issues page

XML Persian Abstract Print

Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:

Alehasher S M J, Teshnehlab M. Implementation of Rough Neural Networks with Probabilistic Learning for Nonlinear System Identification . JoC 2012; 6 (1) :41-50
URL: http://joc.kntu.ac.ir/article-1-67-en.html
Abstract:   (14373 Views)
In this paper an improved rough neural network is presented for identification of chaotic system. Rough neural networks are a type of neural stractures that they are designed based on rough neurons. A rough neuron is considered as a pair of neurons that called lower boandry neuron and upper boandry neuron. Rough neuron approach, allows use of interval computing in neural networks, therefore it can be considered as a new opinion in designing neural networks. The same as multilayer perceptron, rough neural networks also can be trained using by back propagation algorithm based on gradient descending, however, this algorithm has problems such as local minima. In this paper, a new supervised learning method based on effective error of neuron is presented for training of neural networks, which it is called probabilistic learning. To evaluate this study, performance of rough neural network improved, and proposed learning algorithm have been examined in terms of error detection of chaotic time series.
Full-Text [PDF 1311 kb]   (3622 Downloads)    
Type of Article: Research paper | Subject: Special
Received: 2014/06/14 | Accepted: 2014/06/14 | Published: 2014/06/14

Add your comments about this article : Your username or Email:

Send email to the article author

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

© 2023 CC BY-NC 4.0 | Journal of Control

Designed & Developed by : Yektaweb