DocumentCode
2606921
Title
Modified growing neural gas algorithm for faster convergence on signal distribution sudden change
Author
Gancev, Stojanco ; Kulakov, Andrea
Author_Institution
Fac. of Electr. Eng., Univ. of Sts Cyril & Methodius, Skopje, Macedonia
fYear
2009
fDate
29-31 Oct. 2009
Firstpage
1
Lastpage
5
Abstract
The paper deals with the problem of faster optimal coverage of a Growing Neural Gas algorithm for random signals appearing with non-stationary distributions. A modification of the algorithm that successfully solves this problem will be presented with simulations in a 2-D environment and statistical results that will show its efficiency. A comparison with a previous solution for the same problem using so called Utility measure will be also given.
Keywords
neural nets; signal processing; statistical distributions; Utility measure; growing neural gas algorithm; nonstationary distributions; random signal distribution; Convergence; Data mining; Fuzzy neural networks; Fuzzy systems; Size measurement; Sociotechnical systems; Wireless sensor networks; component; faster convergence; fuzzy algorithm; growing neural gas; non-stationary distribution;
fLanguage
English
Publisher
ieee
Conference_Titel
Information, Communication and Automation Technologies, 2009. ICAT 2009. XXII International Symposium on
Conference_Location
Bosnia
Print_ISBN
978-1-4244-4220-1
Electronic_ISBN
978-1-4244-4221-8
Type
conf
DOI
10.1109/ICAT.2009.5348398
Filename
5348398
Link To Document