The Chinese University of Hong Kong

About: The Chinese University of Hong Kong is a education organization based out in Hong Kong, China. It is known for research contribution in the topics: Population & Cancer. The organization has 43411 authors who have published 93672 publications receiving 3066651 citations.

A retrospective on lithium-ion batteries.

Abstract: The 2019 Nobel Prize in Chemistry has been awarded to John B. Goodenough, M. Stanley Whittingham and Akira Yoshino for their contributions in the development of lithium-ion batteries, a technology that has revolutionized our way of life. Here we look back at the milestone discoveries that have shaped the modern lithium-ion batteries for inspirational insights to guide future breakthroughs.

Controlling the quantum stereodynamics of ultracold bimolecular reactions

Abstract: Molecular collisions in the quantum regime represent a new opportunity to explore chemical reactions. Recently, atom-exchange reactions were observed in a trapped ultracold gas of KRb molecules. In an external electric field, these polar molecules can easily be oriented and the exothermic and barrierless bimolecular reactions, KRbC KRb ! K2C Rb2, occur at a rate that rises steeply with increasing dipole moment. Here we demonstrate the suppression of the bimolecular chemical reaction rate by nearly two orders of magnitude when we use an optical lattice trap to confine the fermionic polar molecules in a quasi-two-dimensional, pancake-like geometry, with the dipoles oriented along the tight confinement direction. With the combination of sufficiently tight confinement and Fermi statistics of the molecules, two polar molecules can approach each other only in a ‘side-by-side’ collision under repulsive dipole‐dipole interactions. The suppression of chemical reactions is a prerequisite for the realization of new molecule-based quantum systems.

Monitoring community responses to the SARS epidemic in Hong Kong: from day 10 to day 62

Abstract: Study Objective: To report the evolution in perceptions and behaviours of the general public in response to the severe acute respiratory syndrome (SARS) epidemic in Hong Kong. Design: Ten similar and sequential telephone surveys were conducted during outbreak of SARS, which are classified as belonging to the first and second phases of the epidemic. Setting: Hong Kong, China. Participants: 1397 Hong Kong residents between 18 and 60 years of age. Main outcome measures: Perceptions and behaviours to SARS and its prevention. Results: Most of the respondents believed that SARS could be transmitted via direct body contact and droplets. About half of respondents believed that SARS was curable, which increased in the initial phase and decreased in the second phase. Perceived chance of infection was low (9%) but fear of infection in public places was high (48%). Perceived efficacy of hygiene measures (wearing a mask: 82%, hand washing: 93%, and home disinfection: 75%) remained high in both phases and the perceived efficacy of avoiding crowded place, and using public transportation, etc, increased initially and decreased in the second phase. In parallel, use of the three hygiene measures increased significantly in the first phase and remained high for wearing a mask and washing hands in the second phase. Percentages of people avoiding crowded place and public transportation significantly increased initially and decreased in the second phase. Conclusion: SARS related perceptions and behaviours evolved rapidly during the epidemic and Hong Kong residents quickly adopted appropriate SARS prevention measures. Timely dissemination of information seems effective in public health crises management.

PU-Net: Point Cloud Upsampling Network

Abstract: Learning and analyzing 3D point clouds with deep networks is challenging due to the sparseness and irregularity of the data. In this paper, we present a data-driven point cloud upsampling technique. The key idea is to learn multi-level features per point and expand the point set via a multi-branch convolution unit implicitly in feature space. The expanded feature is then split to a multitude of features, which are then reconstructed to an upsampled point set. Our network is applied at a patch-level, with a joint loss function that encourages the upsampled points to remain on the underlying surface with a uniform distribution. We conduct various experiments using synthesis and scan data to evaluate our method and demonstrate its superiority over some baseline methods and an optimization-based method. Results show that our upsampled points have better uniformity and are located closer to the underlying surfaces.

DeepID-Net: Deformable deep convolutional neural networks for object detection

Abstract: In this paper, we propose deformable deep convolutional neural networks for generic object detection. This new deep learning object detection framework has innovations in multiple aspects. In the proposed new deep architecture, a new deformation constrained pooling (def-pooling) layer models the deformation of object parts with geometric constraint and penalty. A new pre-training strategy is proposed to learn feature representations more suitable for the object detection task and with good generalization capability. By changing the net structures, training strategies, adding and removing some key components in the detection pipeline, a set of models with large diversity are obtained, which significantly improves the effectiveness of model averaging. The proposed approach improves the mean averaged precision obtained by RCNN [14], which was the state-of-the-art, from 31% to 50.3% on the ILSVRC2014 detection test set. It also outperforms the winner of ILSVRC2014, GoogLeNet, by 6.1%. Detailed component-wise analysis is also provided through extensive experimental evaluation, which provide a global view for people to understand the deep learning object detection pipeline.