no code implementations • 22 Jan 2024 • Hong Wei, James Xiao, Yichao Zhang, Xia Hong
This suggests that, in order to account for spatial correlation between pixels, a feature vector associated with each pixel may be a vectorized tensor representing the multiple bands and a local patch as appropriate.
no code implementations • 7 Nov 2023 • Jason Holmes, Rui Peng, Yiwei Li, Jinyu Hu, Zhengliang Liu, Zihao Wu, Huan Zhao, Xi Jiang, Wei Liu, Hong Wei, Jie Zou, Tianming Liu, Yi Shao
IMPORTANCE The response effectiveness of different large language models (LLMs) and various individuals, including medical students, graduate students, and practicing physicians, in pediatric ophthalmology consultations, has not been clearly established yet.
no code implementations • 7 Nov 2023 • Jason Holmes, Shuyuan Ye, Yiwei Li, Shi-Nan Wu, Zhengliang Liu, Zihao Wu, Jinyu Hu, Huan Zhao, Xi Jiang, Wei Liu, Hong Wei, Jie Zou, Tianming Liu, Yi Shao
Methods: A 100-item ophthalmology single-choice test was administered to three different LLMs (GPT-3. 5, GPT-4, and PaLM2) and three different professional levels (medical undergraduates, medical masters, and attending physicians), respectively.
no code implementations • 13 Dec 2022 • Hanning Tang, Liusha Yang, Rui Zhou, Jing Liang, Hong Wei, Xuan Wang, Qingjiang Shi, Zhi-Quan Luo
Using artificial intelligent (AI) to re-design and enhance the current wireless communication system is a promising pathway for the future sixth-generation (6G) wireless network.
no code implementations • 10 Nov 2022 • Brian Ondov, Harsh B. Patel, Ai-Te Kuo, Hanan Samet, John Kastner, Yunheng Han, Hong Wei, Niklas Elmqvist
While participants preferred using the latter two dashboards to perform queries with only a geospatial component or only a temporal component, participants uniformly preferred CoronaViz for queries with both spatial and temporal components, highlighting the utility of a unified spatiotemporal encoding.
1 code implementation • ICLR 2022 • Pengcheng Yang, XiaoMing Zhang, Wenpeng Zhang, Ming Yang, Hong Wei
The recent trend of using large-scale deep neural networks (DNN) to boost performance has propelled the development of the parallel pipelining technique for efficient DNN training, which has resulted in the development of several prominent pipelines such as GPipe, PipeDream, and PipeDream-2BW.
no code implementations • 28 Feb 2020 • John Kastner, Hanan Samet, Hong Wei
With the rapid continuing spread of COVID-19, it is clearly important to be able to track the progress of the virus over time in order to be better prepared to anticipate its emergence and spread in new regions as well as declines in its presence in regions thereby leading to or justifying "reopening" decisions.
no code implementations • 19 Dec 2018 • Raphael Tang, Gefei Yang, Hong Wei, Yajie Mao, Ferhan Ture, Jimmy Lin
Voice-enabled commercial products are ubiquitous, typically enabled by lightweight on-device keyword spotting (KWS) and full automatic speech recognition (ASR) in the cloud.
Automatic Speech Recognition Automatic Speech Recognition (ASR) +4