Naslov
Connectionistic Model of Language Acquisition
(Connectionistic Model Of Language Acquisition)

Autori
Mario, Slugan ; Bojana, Dalbelo-Bašić

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proceedings of the European Modelling Symposium 2006 / John Pollard - London : Delhi : University College London, 2006, 204-207

Skup
European Modelling Symposium 2006

Mjesto i datum
London, Ujedinjeno Kraljevstvo, 11.09.2006. - 12.09.2006

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
language acquisition; word segmentation problem; transitional probability; neural networks

Sažetak
One of the problems of the earliest stages of language acquisition is the word segmentation problem. The word segmentation problem explores the way in which infants learn to break down uninterrupted speech stream into word-equivalent units. One of the proposed classes of solutions to the problem is the distributional solution class ; specifically the use of transitional probabilities. We present a model of a distributional solution to the problem using a neural network. The model shows a strong correlation between the measure of transitional probability and that of mean standard error monitored across all output units. The results coincide with the existing empirical and experimental data.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo

POVEZANOST RADA