- Human Pose and Action Recognition
- Sentiment Analysis and Opinion Mining
- Topic Modeling
- Complex Network Analysis Techniques
- Advanced Text Analysis Techniques
- Data Management and Algorithms
- Text and Document Classification Technologies
- Hand Gesture Recognition Systems
- Catalytic C–H Functionalization Methods
- Advanced biosensing and bioanalysis techniques
- Sulfur-Based Synthesis Techniques
- Molecular Sensors and Ion Detection
- Nanoplatforms for cancer theranostics
- Luminescence and Fluorescent Materials
- Video Analysis and Summarization
- Renal Diseases and Glomerulopathies
- Synthesis and Catalytic Reactions
- Protein Degradation and Inhibitors
- Chemical Synthesis and Reactions
- Anomaly Detection Techniques and Applications
- Advanced Vision and Imaging
- Natural Language Processing Techniques
- Sulfur Compounds in Biology
- Histone Deacetylase Inhibitors Research
- Speech Recognition and Synthesis
Shandong University
2021-2025
National Chin-Yi University of Technology
2024
City University of Hong Kong
2014-2022
Amazon (Germany)
2020-2022
City University of Hong Kong, Shenzhen Research Institute
2016-2021
Kowloon Hospital
2014-2020
Amazon (United States)
2020
Baidu (China)
2017
University of Science and Technology of China
2017
Northwestern University
2011-2016
We present DocFormer - a multi-modal transformer based architecture for the task of Visual Document Understanding (VDU). VDU is challenging problem which aims to understand documents in their varied formats (forms, receipts etc.) and layouts. In addition, pre-trained an unsupervised fashion using carefully designed tasks encourage interaction. uses text, vision spatial features combines them novel self-attention layer. also shares learned embeddings across modalities makes it easy model...
A facile, efficient, and practical method for copper-catalyzed direct C-H amination of benzoxazoles with formamides or secondary amines has been developed. The system can be performed in the absence external base only requires O(2) even air as oxidant. variety substituted benzoxazol-2-amines were synthesized moderate to excellent yield.
A stimuli-responsive drug delivery system (DDS) with bioactive surface is constructed by end-capping mesoporous silica nanoparticles (MSNs) functional peptide-coated gold (GNPs). MSNs are first functionalized acid-labile α-amide-β-carboxyl groups to carry negative charges, and then capped positively charged GNPs that decorated oligo-lysine-containing peptide. The resulting hybrid exhibits endo/lysosomal pH triggered release, the incorporation of RGD peptide facilitates targeting αvβ3...
Although tissue-resident-memory T (TRM) cells, a recently identified non-circulating memory cell population, play crucial role in mediating local immune responses and protect against pathogens upon reinfection, the composition, effector function, specificity of TRM cells kidney their relevance for chronic disease remain unknown. In this study, we found that renal tissue displayed high abundance tissue-resident lymphocytes, proportion CD8+ was significantly increased from patients mice with...
Histone deacetylases (HDACs) play important roles in regulating various physiological and pathological processes. Developing fluorescent probes capable of detecting HDAC activity can help further elucidate the HDACs biology. In this study, we first developed a set activity-based by incorporating Kac residue O-NBD group. Upon enzymatic removal acetyl group residue, released free amine reacted intramolecularly with moiety, resulting turn-on fluorescence. These designed are continuous fashion,...
Although macrophages are undoubtedly attractive therapeutic targets for acute kidney injury (AKI) because of their critical roles in renal inflammation and repair, the underlying mechanisms macrophage phenotype switching efferocytosis regulation inflammatory responses during AKI still largely unclear. The present study elucidated role junctional adhesion molecule–like protein (JAML) pathogenesis AKI. We found that JAML was significantly upregulated kidneys from 2 different murine models...
Positively charged water-soluble polythiophene (PT0) that could self-assemble into nanoparticles in pure water solution was designed and synthesized. PT0 exhibited high photostabilities pH stabilities, excellent biocompatibility, strong 1O2 generation capability, large two-photon absorption cross sections. Moreover, we showed the fluorescence of unaffected by interference biomolecules metal ions. As an example application, demonstrated to be capable simultaneous cell imaging photodynamic...
Lysine lipoylation, a highly conserved lysine post-translational modification, plays critical role in regulating cell metabolism. The catalytic activity of number vital metabolic proteins, such as pyruvate dehydrogenase (PDH), depends on lipoylation. Despite its important roles, the detailed biological regulatory mechanism lipoylation remains largely unexplored. Herein we designed powerful affinity-based probe, KPlip, to interrogate interactions lipoylated peptide/proteins under native...
Social Media is becoming major and popular technological platform that allows users discussing sharing information. Information generated managed through either computer or mobile devices by one person consumed many other persons. Most of these user content are textual information, as Networks (Face book, Linked In), Microblogging (Twitter), blogs (Blogspot, Word press). Looking for valuable nuggets knowledge, such capturing summarizing sentiments from huge amount data could help make...
A series of new fluorogenic probes for monoamine oxidases (MAOs) were reported based on an oxidation and β-elimination mechanism. The limits detection the MAO-A -B determined to be 3.5 6.0 μg mL−1 respectively. These displayed strong activity towards MAOs, especially MAO-B. Cellular imaging studies also successfully conducted with MCF-7 cells.
Estimating 3D hand poses from RGB images is essential to a wide range of potential applications, but challenging owing substantial ambiguity in the inference depth information images. State-of-the-art estimators address this problem by regularizing pose estimation models during training enforce consistency between predicted and ground-truth maps. However, these rely on both paired maps training. In study, we propose conditional generative adversarial network (GAN) model, called Depth-image...
Estimating the 2D human poses in each view is typically first step calibrated multi-view 3D pose estimation. But performance of detectors suffers from challenging situations such as occlusions and oblique viewing angles. To address these challenges, previous works derive point-to-point correspondences between different views epipolar geometry utilize to merge prediction heatmaps or feature representations. Instead post-prediction merge/calibration, here we introduce a transformer framework...
Estimating 3D hand poses from a single RGB image is challenging because depth ambiguity leads the problem ill-posed. Training pose estimators with mesh annotations and multi-view images often results in significant performance gains. However, existing datasets are relatively small joints annotated by off-the-shelf trackers or automated through model predictions, both of which may be inaccurate can introduce biases. Collecting large-scale accurate joint valuable but strenuous. In this paper,...
Although the relationship between macroautophagy/autophagy and Alzheimer disease (AD) is widely studied, underlying mechanisms are poorly understood, especially regulatory role of initiation signaling autophagy on AD. Here, we find that ER transmembrane protein CANX (calnexin) a novel interaction partner autophagy-inducing kinase ULK1 required for recruitment to under basal or starved conditions. Loss results in inactivity inhibits flux. In brains people with AD APP-PSEN1 mice, declines....
Glycosyl sulfonamides have been identified as novel inhibitors of carbonic anhydrase, showing great potential for cancer treatment. However, previous approaches to synthesizing glycosyl require multi steps, resulting in low...
Emerging evidence indicates that lipid droplets (LDs) play important roles in metabolism, energy homeostasis, and cell stress management. Notably, dysregulation of LDs is tightly linked to numerous diseases, including lipodystrophies, cancer, obesity, atherosclerosis, others. The pivotal physiological have led an exploration research recent years. functions are inherently connected the composition their proteome. Current methods for profiling LD proteins mostly utilize fractionation, those...
Large Language Models (LLMs) have revolutionized natural language processing tasks. However, their practical application is constrained by substantial memory and computational demands. Post-training quantization (PTQ) considered an effective method to accelerate LLM inference. Despite its growing popularity in model compression, PTQ deployment faces two major challenges. First, low-bit leads performance degradation. Second, restricted the limited integer computing unit type on GPUs,...
A multilingual sentiment identification system (MuSES) implements three different algorithms. The first algorithm augments previous compositional semantic rules by adding specific to social media. second defines a scoring function that measures the degree of sentiment, instead simply classifying into binary polarities. All such scores are calculated based on large volume customer reviews. Due special characteristics media texts, third takes emoticons, negation word position, and...
A rotor-based fluorescent probe to label exosomes in living cells.
Hydrogen peroxide (H2O2) plays a significant role in regulating variety of biological processes. Dysregulation H2O2 can lead to various diseases. Although numerous fluorescent imaging probes for have been reported, the development ratiometric probe with large Stokes shift remains rather limited. Such shown distinct advantages, such as minimized interference from environment and improved signal-to noise ratio. In this work, we reported new pyrene-based compound Py-VPB vitro. The demonstrated...