Task adaptation using map estimation in N-gram language modeling

Hirokazu Masataki, Yoshinori Sagisaka, Kazuya Hisaki, Tatsuya Kawahara

Research output: Chapter in Book/Report/Conference proceedingConference contribution

21 Citations (Scopus)

Abstract

This paper describes a method of task adaptation in N-gram language modeling, for accurately estimating the N-gram statistics from the small amount of data of the target task. Assuming a task-independent N-gram to be a-priori knowledge, the N-gram is adapted to a target task by MAP (maximum a-posteriori probability) estimation. Experimental results showed that the perplexities of the task adapted models were 15% (trigram), 24% (bigram) lower than those of the task-independent model, and that the perplexity reduction of the adaptation went up to 39% at maximum when the amount of text data in the adapted task was very small.

Original languageEnglish
Title of host publicationICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Editors Anon
PublisherIEEE
Pages783-786
Number of pages4
Volume2
Publication statusPublished - 1997
Externally publishedYes
EventProceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP. Part 1 (of 5) - Munich, Ger
Duration: 1997 Apr 211997 Apr 24

Other

OtherProceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP. Part 1 (of 5)
CityMunich, Ger
Period97/4/2197/4/24

Fingerprint

Statistics
estimating
statistics

ASJC Scopus subject areas

  • Signal Processing
  • Electrical and Electronic Engineering
  • Acoustics and Ultrasonics

Cite this

Masataki, H., Sagisaka, Y., Hisaki, K., & Kawahara, T. (1997). Task adaptation using map estimation in N-gram language modeling. In Anon (Ed.), ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings (Vol. 2, pp. 783-786). IEEE.

Task adaptation using map estimation in N-gram language modeling. / Masataki, Hirokazu; Sagisaka, Yoshinori; Hisaki, Kazuya; Kawahara, Tatsuya.

ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. ed. / Anon. Vol. 2 IEEE, 1997. p. 783-786.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Masataki, H, Sagisaka, Y, Hisaki, K & Kawahara, T 1997, Task adaptation using map estimation in N-gram language modeling. in Anon (ed.), ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. vol. 2, IEEE, pp. 783-786, Proceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP. Part 1 (of 5), Munich, Ger, 97/4/21.
Masataki H, Sagisaka Y, Hisaki K, Kawahara T. Task adaptation using map estimation in N-gram language modeling. In Anon, editor, ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. Vol. 2. IEEE. 1997. p. 783-786
Masataki, Hirokazu ; Sagisaka, Yoshinori ; Hisaki, Kazuya ; Kawahara, Tatsuya. / Task adaptation using map estimation in N-gram language modeling. ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. editor / Anon. Vol. 2 IEEE, 1997. pp. 783-786
@inproceedings{9a007d2ec06c467884f52fc051bb07b5,
title = "Task adaptation using map estimation in N-gram language modeling",
abstract = "This paper describes a method of task adaptation in N-gram language modeling, for accurately estimating the N-gram statistics from the small amount of data of the target task. Assuming a task-independent N-gram to be a-priori knowledge, the N-gram is adapted to a target task by MAP (maximum a-posteriori probability) estimation. Experimental results showed that the perplexities of the task adapted models were 15{\%} (trigram), 24{\%} (bigram) lower than those of the task-independent model, and that the perplexity reduction of the adaptation went up to 39{\%} at maximum when the amount of text data in the adapted task was very small.",
author = "Hirokazu Masataki and Yoshinori Sagisaka and Kazuya Hisaki and Tatsuya Kawahara",
year = "1997",
language = "English",
volume = "2",
pages = "783--786",
editor = "Anon",
booktitle = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",
publisher = "IEEE",

}

TY - GEN

T1 - Task adaptation using map estimation in N-gram language modeling

AU - Masataki, Hirokazu

AU - Sagisaka, Yoshinori

AU - Hisaki, Kazuya

AU - Kawahara, Tatsuya

PY - 1997

Y1 - 1997

N2 - This paper describes a method of task adaptation in N-gram language modeling, for accurately estimating the N-gram statistics from the small amount of data of the target task. Assuming a task-independent N-gram to be a-priori knowledge, the N-gram is adapted to a target task by MAP (maximum a-posteriori probability) estimation. Experimental results showed that the perplexities of the task adapted models were 15% (trigram), 24% (bigram) lower than those of the task-independent model, and that the perplexity reduction of the adaptation went up to 39% at maximum when the amount of text data in the adapted task was very small.

AB - This paper describes a method of task adaptation in N-gram language modeling, for accurately estimating the N-gram statistics from the small amount of data of the target task. Assuming a task-independent N-gram to be a-priori knowledge, the N-gram is adapted to a target task by MAP (maximum a-posteriori probability) estimation. Experimental results showed that the perplexities of the task adapted models were 15% (trigram), 24% (bigram) lower than those of the task-independent model, and that the perplexity reduction of the adaptation went up to 39% at maximum when the amount of text data in the adapted task was very small.

UR - http://www.scopus.com/inward/record.url?scp=0030715922&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0030715922&partnerID=8YFLogxK

M3 - Conference contribution

VL - 2

SP - 783

EP - 786

BT - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

A2 - Anon, null

PB - IEEE

ER -