Graphical model architectures for speech recognition

被引：78

作者：

Bilmes, JA

Bartels, C

机构：

[1] Department of Electrical Engineering, University of Washington, Seattle, WA

来源：

IEEE SIGNAL PROCESSING MAGAZINE | 2005年 / 22卷 / 05期

基金：

美国国家科学基金会;

关键词：

D O I：

10.1109/MSP.2005.1511827

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

Statistical graphical models are a family of graphical abstractions of statistical models where the important aspects of such models are represented using graphs. Their main advantage is rapidity that quickly expresses a novel, complicated idea in an intuitive, concise and mathematically formal but widely flexible means of solving problems in speech and language processing. By implementing these models, researchers are able to experiment quickly, reject ideas that perform poorly, and advance ideas that perform well.

引用

页码：89 / 100

页数：12

共 51 条

[1]

[Anonymous], 1999, Probabilistic Networks and Expert Systems

[2]

BARTELS C, 2004, UWEETR20040010

[3]

BEAL MJ, 2002, P 7 EUR C COMP VIS, P736

[4]

Bilmes J, 2002, INT CONF ACOUST SPEE, P3916

[5]

Bilmes J, 1996, ICSLP 96 - FOURTH INTERNATIONAL CONFERENCE ON SPOKEN LANGUAGE PROCESSING, PROCEEDINGS, VOLS 1-4, P1301, DOI 10.1109/ICSLP.1996.607851

[6]

BILMES J, 2001, JHU WS 2001 FIN WORK

[7]

Bilmes J., 2003, UNCERTAINTY ARTIFICI, P47

[8]

Bilmes J., 2000, P 16 C UNC ART INT, P38

[9]

BILMES J, 2004, KUWEETR20040016

[10] Buried Markov models: a graphical-modeling approach to automatic speech recognition [J].

Bilmes, JA .

COMPUTER SPEECH AND LANGUAGE, 2003, 17 (2-3) :213-231

← 1 2 3 4 5 6 →