| Document Date: 2013-10-24 15:15:32
Open Document File Size: 258,29 KBShare Result on Facebook
Company IBM / Google / YouTube / Microsoft Research / IBM Research / DBN-DNN / MFCCs / Deep Neural Networks / heteroscedastic LDA / / Currency USD / / Event Reorganization / / Facility National Institute of Standards and Technology / University of Toronto / / IndustryTerm speech recognition systems / grid search / bottleneck systems / above learning algorithm / machine learning algorithms / frame-based training algorithm / mobile voice input applications / neural networks / approximate learning algorithm / shallow neural networks / deep belief network / artificial neural network / e -x / final network / mobile voice search application / Artificial neural networks / energy / voice search queries / ost current speech recognition systems / feed-forward neural network / energy function / forward-backward algorithm / filter-bank coefficients / neural network / baseline systems / filter-bank outputs / voice search task / filter-bank features / voice search data / speech processing / search queries / deeper networks / computing / hybrid systems / log filter-bank / learning algorithms / using neural networks / / Organization National Institute of Standards and Technology / University of Toronto / / Person George Dahl / Vincent Vanhoucke / Geoffrey Hinton / Dong Yu / Tara Sainath / Patrick Nguyen / Brian Kingsbury / Navdeep Jaitly / / Position speaker / feed-forward / Fisher / / PublishedMedium the IEEE SIGNAL PROCESSING MAGAZINE / / RadioStation DNN / / Technology approximate learning algorithm / Speech Recognition / machine learning algorithms / EM algorithm / mobile devices / frame-based training algorithm / cellular telephone / neural network / above learning algorithm / learning algorithms / Digital Object Identifier / t - 1 / forward-backward algorithm / learned using the EM algorithm / /
SocialTag |
IEEE-28MSP01-Feature-TEMPLATE