Článek ve sborníku konference

 
Karafiát, M., Szőke, I., Černocký, J.: Using Gradient Descent Optimization for Acoustic Training from Heterogeneous Data, In: Proc. Text, Speech and Dialog 2010, Brno, CZ, Springer, 2010, s. 322-329, ISBN 978-3-642-15759-2, ISSN 0302-9743
Jazyk publikace:angličtina
Název publikace:Using Gradient Descent Optimization for Acoustic Training from Heterogeneous Data
Název (cs):Využití gradient descent optimalizace pro trénování akustických modelů z heterogenních dat
Strany:322-329
Sborník:Proc. Text, Speech and Dialog 2010
Konference:13th International Conference on Text, Speech and Dialogue, TSD 2010
Řada knih:LNAI 6231
Místo vydání:Brno, CZ
Rok:2010
ISBN:978-3-642-15759-2
Časopis:Lecture Notes in Computer Science, roč. 2010, č. 9, DE
ISSN:0302-9743
Vydavatel:Springer Verlag
URL:http://www.fit.vutbr.cz/research/groups/speech/publi/2010/karafiat_TSD_2010_322.pdf [PDF]
Klíčová slova
speech, acoustic models, heterogeneous data, HLDA system, gradient descent training, robustness
Anotace
Článek pojednává o využití gradient descent otimalizace pro trénovaní akustických modelů z heterogenních dat. Zabýváme se využitím heterogenních dat pro trénování akustických modelů. 
Abstrakt
In this paper, we study the use of heterogeneous data for training of acoustic models. In initial experiments, a significant drop of accuracy has been observed on in-domain test set if the data was added without any regularization. A solution is proposed by getting control over the training data by optimization of the weights of different data-sets. The final models shows good performance on all various tests linked to various speaking styles. Furthermore, we used this approach to increase the performance over just the main test set. We obtained 0.3% absolute improvement on basic system and 0.4% on HLDA system although the size of the heterogeneous data set was quite small.
BibTeX:
@INPROCEEDINGS{
   author = {Martin Karafiát and Igor Szőke and Jan Černocký},
   title = {Using Gradient Descent Optimization for Acoustic Training
	from Heterogeneous Data},
   pages = {322--329},
   booktitle = {Proc. Text, Speech and Dialog 2010},
   series = {LNAI 6231},
   journal = {Lecture Notes in Computer Science},
   volume = {2010},
   number = {9},
   year = {2010},
   location = {Brno, CZ},
   publisher = {Springer Verlag},
   ISBN = {978-3-642-15759-2},
   ISSN = {0302-9743},
   language = {english},
   url = {http://www.fit.vutbr.cz/research/view_pub.php?id=9322}
}