Výsledky vyhledávání - "Herbert Gish"

Towards a New Understanding of the Training of Neural Networks with Mislabeled Training Data

Autor: Man-Hung Siu, William Hartmann, Jan Silovsky, Zhuolin Jiang, Herbert Gish, Man-Ling Sung

Publikováno v: ICASSP

We investigate the problem of machine learning with mislabeled training data. We try to make the effects of mislabeled training better understood through analysis of the basic model and equations that characterize the problem. This includes results a

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::99c94e101a911902a2cba2c398dcb32b
http://arxiv.org/abs/1909.09136

Zobrazit plný text záznamu

Unsupervised training of an HMM-based self-organizing unit recognizer with applications to topic classification and keyword discovery

Autor: William Belfield, Arthur Chan, Steve Lowe, Man-Hung Siu, Herbert Gish

Publikováno v: Computer Speech & Language. 28:210-223

We present our approach to unsupervised training of speech recognizers. Our approach iteratively adjusts sound units that are optimized for the acoustic domain of interest. We thus enable the use of speech recognizers for applications in speech domai

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_________::e8f4760fd9869eed65aee73ba6a6600b
https://doi.org/10.1016/j.csl.2013.05.002

Zobrazit plný text záznamu

Importance sampling of delta-AUC: A basis for active learning for improved keyword search

Autor: Kerri Barnes, Matthew Snover, Man-Hung Siu, Herbert Gish

Publikováno v: ICASSP

We present an importance sampling based approach to the active learning problem of selecting additional training data to supplement a seed model. Our proposed Δ-AUC selection optimizes AUC improvement in keyword search and is evaluated on the Spanis

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_________::ae713a9c848a4aec14a8784b9ac55aa6
https://doi.org/10.1109/icassp.2016.7472076

Zobrazit plný text záznamu

Discriminatively Trained GMMs for Language Classification Using Boosting Methods

Autor: Herbert Gish, Xi Yang, Man-Hung Siu

Publikováno v: IEEE Transactions on Audio, Speech, and Language Processing. 17:187-197

In language identification and other speech applications, discriminatively trained models often outperform nondiscriminative models trained with the maximum-likelihood criterion. For instance, discriminative Gaussian mixture models (GMMs) are typical

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_________::3b46a053a4f8903b487d794bb090f6ce
https://doi.org/10.1109/tasl.2008.2006653

Zobrazit plný text záznamu

Evaluation of word confidence for speech recognition systems

Autor: Herbert Gish, Man-Hung Siu

Publikováno v: Computer Speech & Language. 13:299-319

Confidence measures enable us to assess the output of a speech recognition system. The confidence measure provides us with an estimate of the probability that a word in the recognizer output is either correct or incorrect. In this paper we discuss wa

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_________::da96d1344ca888f451b968b97d1d2d8f
https://doi.org/10.1006/csla.1999.0126

Zobrazit plný text záznamu

MLLR transforms of self-organized units as features in speaker recognition

Autor: Omer Lang, Herbert Gish, Arthur Chan, Owen Kimball, Man-Hung Siu, Steve Lowe

Publikováno v: ICASSP

Using speaker adaptation parameters, such as maximum likelihood linear regression (MLLR) adaptation matrices, as features for speaker recognition (SR) has been shown to perform well and can also provide complementary information for fusion with other

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_________::7c3e232a998735d0784e49a40e6e3460
https://doi.org/10.1109/icassp.2012.6288891

Zobrazit plný text záznamu

Topic modeling for spoken documents using only phonetic information

Autor: Arthur Chan, Man-Hung Siu, Timothy J. Hazen, Herbert Gish, Steve Lowe

Publikováno v: ASRU

This paper explores both supervised and unsupervised topic modeling for spoken audio documents using only phonetic information. In cases where word-based recognition is unavailable or infeasible, phonetic information can be used to indirectly learn a

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_________::08075687e8335809af4919ebf1442d88
https://doi.org/10.1109/asru.2011.6163964

Zobrazit plný text záznamu

Vyhledávací nástroje:

Upřesnit hledání