Pregled bibliografske jedinice broj: 5783
An approach to synthesis of adaptive recurrent neural networks for robot control
An approach to synthesis of adaptive recurrent neural networks for robot control // Proceedings-KoREMA"96-41st Annual Conference, Vol. 2 / Vukic, Zoran (ur.).
Zagreb: Hrvatsko društvo za komunikacije, računarstvo, elektroniku, mjerenja I automatiku (KoREMA), 1996. str. 125-128 (predavanje, međunarodna recenzija, cjeloviti rad (in extenso), znanstveni)
CROSBI ID: 5783 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
An approach to synthesis of adaptive recurrent neural networks for robot control
Autori
Novaković, Branko ; Majetić, Dubravko
Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni
Izvornik
Proceedings-KoREMA"96-41st Annual Conference, Vol. 2
/ Vukic, Zoran - Zagreb : Hrvatsko društvo za komunikacije, računarstvo, elektroniku, mjerenja I automatiku (KoREMA), 1996, 125-128
Skup
KoREMA"96-41st Annual Conference
Mjesto i datum
Opatija, Hrvatska, 18.09.1996. - 20.09.1996
Vrsta sudjelovanja
Predavanje
Vrsta recenzije
Međunarodna recenzija
Ključne riječi
Recurrent neural network; adaptation; robot control
Sažetak
Abstract - In this paper a new algorithm for synthesis of adaptive recurrent discrete-time neural networks (NN) is proposed. For this purpose the following concepts are employed: (i) time-varying NN weights distribution in interaction weight matrix, (ii) one-step learning iteration approach in off-line learning of recurrent NN (RNN), and (iii) adaptation of RNN during the NN working phase by employing a new structure of a dynamic back-propagation neural network in on-line learning. Following the proposed NN synthesis procedure the adaptive recurrent NN for an adaptive nonlinear robot control is designed.
Izvorni jezik
Engleski
Znanstvena područja
Strojarstvo
