Zobrazeno 1 - 10
of 337
pro vyhledávání: '"Rao, Ma"'
Combining serum AFP and CEUS LI-RADS for better diagnostic performance in Chinese high-risk patients
Autor:
Wushuang Gong, Jiaqi Wu, Hong Wei, Zhaopeng Jiang, Ming Wan, Chengwei Wu, Weili Xue, Rao Ma, Xianli Zhou, Hang Zhou
Publikováno v:
La radiologia medica. 128:393-401
Publikováno v:
Psychology Research and Behavior Management.
Mengyang Zhao,1 Liu Yao,2 Rao Ma,3 Muhammad Sarmad,4 Orangzab,5 Arslan Ayub,6 Zhou Jun7 1School of Economics and Management, Harbin Engineering University, Harbin, Heilongjing, 150001, Peopleâs Republic of China; 2Faculty of Management Engineeri
Publikováno v:
2022 International Conference on Education, Network and Information Technology (ICENIT).
Publikováno v:
IEEE/ACM Transactions on Audio, Speech, and Language Processing. 28:2438-2449
Large memory consumption of the neural network language models (NN LMs) prohibits their use in many resource-constrained scenarios. Hence, effective NN LM compression approaches that are independent of NN structures are of great interest. However, pr
An end-to-end (E2E) ASR model implicitly learns a prior Internal Language Model (ILM) from the training transcripts. To fuse an external LM using Bayes posterior theory, the log likelihood produced by the ILM has to be accurately estimated and subtra
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::977f9a9d1b83da3c479dd59dd8f3d3ba
Publikováno v:
Sustainability; Volume 14; Issue 24; Pages: 16571
High-quality innovation can solve the “bottleneck” problem of key enterprise technologies and drive the high-quality development of enterprises. Therefore, how to improve innovation quality has become a growing concern in the academic industry. I
Publikováno v:
ICASSP
This paper describes the AISpeech-SJTU ASR system for the Interspeech-2020 Accented English Speech Recognition Challenge (AESRC). This task is challenging due to the diversity of pronunciation accuracy, intonation speed and pronunciation of some syll
Publikováno v:
ICASSP
This paper describes the AISpeech-SJTU system for the accent identification track of the Interspeech-2020 Accented English Speech Recognition Challenge. In this challenge track, only 160-hour accented English data collected from 8 countries and the a
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::dae91e6747774c8242c59a0171e75ea9
http://arxiv.org/abs/2102.09828
http://arxiv.org/abs/2102.09828
Publikováno v:
ACL
One daunting problem for semantic parsing is the scarcity of annotation. Aiming to reduce nontrivial human labor, we propose a two-stage semantic parsing framework, where the first stage utilizes an unsupervised paraphrase model to convert an unlabel
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::2cda292af064c88df19131834ed5078e
http://arxiv.org/abs/2005.13485
http://arxiv.org/abs/2005.13485
Publikováno v:
ICASSP
Neural network language models have gained considerable popularity due to their promising performance. Distributed word embeddings are utilized to represent semantic information. However, each word is associated with a single vector in the embedding