Článek ve sborníku konference

GRÉZL František, EGOROVA Ekaterina a KARAFIÁT Martin. Study of Large Data Resources for Multilingual Training and System Porting. In: Procedia Computer Science. Yogyakarta: Elsevier Science, 2016, s. 15-22. ISSN 1877-0509. Dostupné z: http://www.sciencedirect.com/science/article/pii/S1877050916300382
Jazyk publikace:angličtina
Název publikace:Study of Large Data Resources for Multilingual Training and System Porting
Název (cs):Studie velkých datových zdrojů pro multilingvální trénování a portování systémů
Strany:15-22
Sborník:Procedia Computer Science
Konference:The 5th International Workshop on Spoken Language Technologies for Under-resourced Languages (SLTU'16)
Místo vydání:Yogyakarta, ID
Rok:2016
URL:http://www.sciencedirect.com/science/article/pii/S1877050916300382
Časopis:Procedia Computer Science, roč. 2016, č. 81, CZ
ISSN:1877-0509
DOI:10.1016/j.procs.2016.04.024
Vydavatel:Elsevier Science
URL:http://www.fit.vutbr.cz/research/groups/speech/publi/2016/grezl_sltu2016_08-8028.pdf [PDF]
Klíčová slova
Stacked Bottle-Neck; feature extraction; multilingual training; large data; Fisher database
Anotace
Článek pojednává o studii velkých datových zdrojů pro multilinguální trénování a portování systémů. Tato studie využívá velké databáze pro natrénování neuronové sítě pro extrakci příznaků. Ta je následně portována do cílového jazyka pro který jsou k dispozici jen omezená data. U zdrojového jazyka měníme množství dat a velikost akustických jednotek (trifonové shlukování). Následně se vyhodnocuje efekt těchto změn na úspěšnosti systému v cílovém jazyce.
Abstrakt
This study investigates the behavior of a feature extraction neural network model trained on a large amount of single language data ("source language") on a set of under-resourced target languages. The coverage of the source language acoustic space was changed in two ways: (1) by changing the amount of training data and (2) by altering the level of detail of acoustic units (by changing the triphone clustering). We observe the effect of these changes on the performance on target language in two scenarios: (1) the source-language NNs were used directly, (2) NNs were first ported to target language. The results show that increasing coverage as well as level of detail on the source language improves the target language system performance in both scenarios. For the first one, both source language characteristic have about the same effect. For the second scenario, the amount of data in source language is more important than the level of detail. The possibility to include large data into multilingual training set was also investigated. Our experiments point out possible risk of over-weighting the NNs towards the source language with large data. This degrades the performance on part of the target languages, compared to the setting where the amounts of data per language are balanced.
BibTeX:
@INPROCEEDINGS{
   author = {Franti{\v{s}}ek Gr{\'{e}}zl and Ekaterina Egorova
	and Martin Karafi{\'{a}}t},
   title = {Study of Large Data Resources for Multilingual
	Training and System Porting},
   pages = {15--22},
   booktitle = {Procedia Computer Science},
   journal = {Procedia Computer Science},
   volume = 2016,
 number = 81,
   year = 2016,
   location = {Yogyakarta, ID},
   publisher = {Elsevier Science},
   ISSN = {1877-0509},
   doi = {10.1016/j.procs.2016.04.024},
   language = {english},
   url = {http://www.fit.vutbr.cz/research/view_pub.php.cs?id=11222}
}

Vaše IPv4 adresa: 3.81.28.94
Přepnout na https