Task adaptation using MAP estimation in N-gram language modeling

Author

Masataki, Hirokazu ; Sagisaka, Yoshinori ; Hisaki, Kazuya ; Kawahara, Tatsuya

Author_Institution

ATR Interpreting Telephony Res. Labs., Kyoto, Japan

Volume

2

fYear

1997

fDate

21-24 Apr 1997

Firstpage

783

Abstract

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) and 24% (bigram) lower than those of the task-independent model, and that the perplexity reduction of the adaptation went up to a maximum of 39% when the amount of text data in the adapted task was very small

Keywords

maximum likelihood estimation; natural languages; nomograms; speech recognition; statistics; MAP estimation; N-gram language modeling; N-gram statistics estimation; bigram; continuous speech recognition; maximum a-posteriori probability estimation; perplexity reduction; task adaptation; task-independent N-gram; task-independent model; text data; trigram; Equations; Information science; Maximum a posteriori estimation; Maximum likelihood estimation; Natural languages; Parameter estimation; Probability; Smoothing methods; Speech recognition; Statistics;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1997. ICASSP-97., 1997 IEEE International Conference on

Conference_Location

Munich

ISSN

1520-6149

Print_ISBN

0-8186-7919-0

Type

conf

DOI

10.1109/ICASSP.1997.596042

Filename

596042