User profiles for Yujun Shi
Yujun Shi (施宇钧)PhD Student, National University of Singapore Verified email at u.nus.edu Cited by 2614 |
Tokens-to-token vit: Training vision transformers from scratch on imagenet
Transformers, which are popular for language modeling, have been explored for solving
vision tasks recently, eg, the Vision Transformer (ViT) for image classification. The ViT model …
vision tasks recently, eg, the Vision Transformer (ViT) for image classification. The ViT model …
A new perspective on the operating principle of flux-switching permanent-magnet machines
The flux-switching permanent-magnet (FSPM) machine attracts increasing attention recently
due to its high power density, robust mechanical structure, good flux-weakening capability, …
due to its high power density, robust mechanical structure, good flux-weakening capability, …
A novel dual-permanent-magnet-excited machine for low-speed large-torque applications
L Jian, Y Shi, C Liu, G Xu, Y Gong… - IEEE transactions on …, 2013 - ieeexplore.ieee.org
This paper proposes a novel dual-permanent-magnet-excited (DPME) machine, in which
two sets of permanent magnets (PMs) are employed, one on stator and the other on rotor. Bi-…
two sets of permanent magnets (PMs) are employed, one on stator and the other on rotor. Bi-…
All tokens matter: Token labeling for training better vision transformers
In this paper, we present token labeling---a new training objective for training high-performance
vision transformers (ViTs). Different from the standard training objective of ViTs that …
vision transformers (ViTs). Different from the standard training objective of ViTs that …
Smart face mask based on an ultrathin pressure sensor for wireless monitoring of breath conditions
…, D Inoue, D Hashizume, Z Jiang, Y Shi… - Advanced …, 2022 - Wiley Online Library
A smart face mask that can conveniently monitor breath information is beneficial for maintaining
personal health and preventing the spread of diseases. However, some challenges still …
personal health and preventing the spread of diseases. However, some challenges still …
Leveraging instance-, image-and dataset-level information for weakly supervised instance segmentation
Weakly supervised semantic instance segmentation with only image-level supervision,
instead of relying on expensive pixel-wise masks or bounding box annotations, is an important …
instead of relying on expensive pixel-wise masks or bounding box annotations, is an important …
Osteoinduction of hydroxyapatite/β-tricalcium phosphate bioceramics in mice with a fractured fibula
L Cheng, F Ye, R Yang, X Lu, Y Shi, L Li, H Fan, H Bu - Acta biomaterialia, 2010 - Elsevier
Many studies have shown that calcium phosphate ceramics can induce bone formation in
non-osseous sites without the application of any osteoinductive biomolecules, but the …
non-osseous sites without the application of any osteoinductive biomolecules, but the …
Learning with fantasy: Semantic-aware virtual contrastive constraint for few-shot class-incremental learning
Few-shot class-incremental learning (FSCIL) aims at learning to classify new classes
continually from limited samples without forgetting the old classes. The mainstream framework …
continually from limited samples without forgetting the old classes. The mainstream framework …
[HTML][HTML] Exosomes derived from human umbilical cord mesenchymal stem cells ameliorate IL-6-induced acute liver injury through miR-455-3p
…, X Cao, M Chen, B Zhang, Y Zhou, R Yao, Y Shi… - Stem cell research & …, 2020 - Springer
Background Using a toxin-induced nonhuman primate model of acute liver failure (ALF), we
previously reported that peripheral infusion of human umbilical cord mesenchymal stem …
previously reported that peripheral infusion of human umbilical cord mesenchymal stem …
Dragdiffusion: Harnessing diffusion models for interactive point-based image editing
Precise and controllable image editing is a challenging task that has attracted significant
attention. Recently, DragGAN enables an interactive point-based image editing framework and …
attention. Recently, DragGAN enables an interactive point-based image editing framework and …