Článek ve sborníku konference

 
Mikolov, T., Kopecký, J., Burget, L., Glembek, O., Černocký, J.: Neural network based language models for highly inflective languages, In: Proc. ICASSP 2009, Taipei, TW, IEEESP, 2009, s. 4, ISBN 978-1-4244-2354-5
Jazyk publikace:angličtina
Název publikace:Neural network based language models for highly inflective languages
Název (cs):Jazykové modely založené na neuronových sítích pro vysoce ohebné jazyky
Strany:4
Sborník:Proc. ICASSP 2009
Konference:International Conference on Acoustics, Speech, and Signal Processing
Místo vydání:Taipei, TW
Rok:2009
ISBN:978-1-4244-2354-5
Vydavatel:IEEE Signal Processing Society
URL:http://www.fit.vutbr.cz/research/groups/speech/publi/2009/mikolov_ic2009_nnlm_4.pdf [PDF]
Klíčová slova
language modeling, neural networks, inflective languages
Anotace
Článek je o jazykových modelech založených na neuronových sítích pro vysoce ohebné jazyky
Abstrakt
Speech recognition of inflectional and morphologically rich languages like Czech is currently quite a challenging task, because simple n-gram techniques are unable to capture important regularities in the data. Several possible solutions were proposed, namely class based models, factored models, decision trees and neural networks. This paper describes improvements obtained in recognition of spoken Czech lectures using languagemodels based on neural networks. Relative reductions in word error rate are more than 15% over baseline obtained with adapted 4-gram backoff language model using modified Kneser-Ney smoothing.
BibTeX:
@INPROCEEDINGS{
   author = {Tomáš Mikolov and Jiří Kopecký and Lukáš Burget and Ondřej
	Glembek and Jan Černocký},
   title = {Neural network based language models for highly inflective
	languages},
   pages = {4},
   booktitle = {Proc. ICASSP 2009},
   year = {2009},
   location = {Taipei, TW},
   publisher = {IEEE Signal Processing Society},
   ISBN = {978-1-4244-2354-5},
   language = {english},
   url = {http://www.fit.vutbr.cz/research/view_pub.php?id=9034}
}