BackgroundCheck.run
Search For

Jia Cui, 50Bellevue, WA

Jia Cui Phones & Addresses

Bellevue, WA   

Millwood, NY   

Yorktown Heights, NY   

Mount Kisco, NY   

4010 Linkwood Rd, Baltimore, MD 21210    410-8898955   

110 39Th St, Baltimore, MD 21210    410-3668635   

340 Stevenson St, Towson, MD 21204    410-8231185   

Social networks

Jia Cui
Jia Cui

Linkedin

Work

Company: St james's place Jan 2011 to May 2011 Address: London, United Kingdom Position: Investment assistant

Education

Degree: BS School / High School: Penn State University 2008 to 2012 Specialities: Finance, International Business

Languages

English • Chinese

Awards

Penn State President's Freshman Award, Feb. 2010

Industries

Management Consulting
Get Report
Control this profile

Mentions for Jia Cui

Jia Cui resumes & CV records

Resumes

Jia Cui Photo 30

Associate At Jl Consulting

Location:
Shanghai City, China
Industry:
Management Consulting
Work:
St James's Place - London, United Kingdom Jan 2011 - May 2011
Investment Assistant Harristown Development Corporation - Harrisburg, Pennsylvania Area May 2010 - Aug 2010
Vice President Assistant
Education:
Penn State University 2008 - 2012
BS, Finance, International Business
Honor & Awards:
Penn State President's Freshman Award, Feb. 2010
Languages:
English
Chinese

Publications & IP owners

Us Patents

Token-Wise Training For Attention Based End-To-End Speech Recognition

US Patent:
2021026, Aug 26, 2021
Filed:
May 11, 2021
Appl. No.:
17/316856
Inventors:
- Palo Alto CA, US
Jia CUI - Bellevue WA, US
Chao WENG - Fremont CA, US
Dong YU - Bothell WA, US
Assignee:
TENCENT AMERICA LLC - Palo Alto CA
International Classification:
G10L 15/06
G06N 20/00
G10L 15/22
G10L 15/14
G06N 7/00
Abstract:
A method of attention-based end-to-end (A-E2E) automatic speech recognition (ASR) training, includes performing cross-entropy training of a model, based on one or more input features of a speech signal, determining a posterior probability vector at a time of a first wrong token among one or more output tokens of the model of which the cross-entropy training is performed, and determining a loss of the first wrong token at the time, based on the determined posterior probability vector. The method further includes determining a total loss of a training set of the model of which the cross-entropy training is performed, based on the determined loss of the first wrong token, and updating the model of which the cross-entropy training is performed, based on the determined total loss of the training set.

Token-Wise Training For Attention Based End-To-End Speech Recognition

US Patent:
2020026, Aug 20, 2020
Filed:
Feb 14, 2019
Appl. No.:
16/275971
Inventors:
- Palo Alto CA, US
Jia Cui - Bellevue WA, US
Chao Weng - Fremont CA, US
Dong Yu - Bothell WA, US
Assignee:
TENCENT AMERICA LLC - Palo Alto CA
International Classification:
G10L 15/06
G06N 20/00
G06N 7/00
G10L 15/22
G10L 15/14
Abstract:
A method of attention-based end-to-end (A-E2E) automatic speech recognition (ASR) training, includes performing cross-entropy training of a model, based on one or more input features of a speech signal, determining a posterior probability vector at a time of a first wrong token among one or more output tokens of the model of which the cross-entropy training is performed, and determining a loss of the first wrong token at the time, based on the determined posterior probability vector. The method further includes determining a total loss of a training set of the model of which the cross-entropy training is performed, based on the determined loss of the first wrong token, and updating the model of which the cross-entropy training is performed, based on the determined total loss of the training set.

Large Margin Training For Attention-Based End-To-End Speech Recognition

US Patent:
2020026, Aug 20, 2020
Filed:
Feb 14, 2019
Appl. No.:
16/276081
Inventors:
- Palo Alto CA, US
Jia CUI - Bellevue WA, US
Chao WENG - Fremont CA, US
Dong YU - Bothell WA, US
Assignee:
Tencent America LLC - Palo Alto CA
International Classification:
G10L 15/06
G10L 15/30
Abstract:
A method of attention-based end-to-end (EE) automatic speech recognition (ASR) training, includes performing cross-entropy training of a model, based on one or more input features of a speech signal, performing beam searching of the model of which the cross-entropy training is performed, to generate an n-best hypotheses list of output hypotheses, and determining a one-best hypothesis among the generated n-best hypotheses list. The method further includes determining a character-based gradient and a word-based gradient, based on the model of which the cross-entropy training is performed and a loss function in which a distance between a reference sequence and the determined one-best hypothesis is maximized, and performing backpropagation of the determined character-based gradient and the determined word-based gradient to the model, to update the model.

Input-Feeding Architecture For Attention Based End-To-End Speech Recognition

US Patent:
2020011, Apr 16, 2020
Filed:
Oct 15, 2018
Appl. No.:
16/160352
Inventors:
- Palo Alto CA, US
Jia Cui - Bellevue WA, US
Guangsen WANG - Shenzhen, CN
Jun Wang - Shenzhen, CN
Chengzhu Yu - Bellevue WA, US
Dan Su - Shenzhen, CN
Dong Yu - Bothell WA, US
Assignee:
TENCENT AMERICA LLC - Palo Alto CA
International Classification:
G10L 15/06
G10L 15/22
Abstract:
Methods and apparatuses are provided for performing end-to-end speech recognition training performed by at least one processor. The method includes receiving, by the at least one processor, one or more input speech frames, generating, by the at least one processor, a sequence of encoder hidden states by transforming the input speech frames, computing, by the at least one processor, attention weights based on each of the sequence of encoder hidden states and a current decoder hidden state, performing, by the at least one processor, a decoding operation based on a previous embedded label prediction information and a previous attentional hidden state information generated based on the attention weights; and generating a current embedded label prediction information based on a result of the decoding operation and the attention weights.

Multistage Curriculum Training Framework For Acoustic-To-Word Speech Recognition

US Patent:
2020007, Mar 5, 2020
Filed:
Aug 30, 2018
Appl. No.:
16/117373
Inventors:
- Palo Alto CA, US
Chao WENG - Fremont CA, US
Jia CUI - Bellevue WA, US
Dong YU - Bothell WA, US
Assignee:
TENCENT AMERICA LLC - Palo Alto CA
International Classification:
G10L 15/06
G10L 15/16
Abstract:
Methods and apparatuses are provided for performing acoustic to word (A2W) speech recognition training performed by at least one processor. The method includes initializing, by the at least one processor, one or more first layers of a neural network with phone based Connectionist Temporal Classification (CTC), initializing, by the at least one processor, one or more second layers of the neural network with grapheme based CTC, acquiring, by the at least one processor, training data and performing, by the at least one processor, A2W speech recognition training based the initialized one or more first layers and one or more second layers of the neural network using the training data.

Possible Relatives

Daniel T Cui  Dong Cui  Fang D Cui  Feng Cui  Guiying Y Cui  Hua L Cui  James B Cui  Jia Z Li  Jia W Lin  Jia X Liu 

NOTICE: You may not use BackgroundCheck or the information it provides to make decisions about employment, credit, housing or any other purpose that would require Fair Credit Reporting Act (FCRA) compliance. BackgroundCheck is not a Consumer Reporting Agency (CRA) as defined by the FCRA and does not provide consumer reports.