A5100719418- Advanced Neural Network Applications
- Video Surveillance and Tracking Methods
- Advanced Image and Video Retrieval Techniques
- Single-cell and spatial transcriptomics
- Nanoplatforms for cancer theranostics
- Domain Adaptation and Few-Shot Learning
- Advanced Nanomaterials in Catalysis
- Network Traffic and Congestion Control
- Gene expression and cancer classification
- Human Pose and Action Recognition
- Industrial Technology and Control Systems
- Higher Education and Teaching Methods
- Infrared Target Detection Methodologies
- Advanced Image Fusion Techniques
- Visual Attention and Saliency Detection
- Bioinformatics and Genomic Networks
- Advanced Sensor and Energy Harvesting Materials
- Gene Regulatory Network Analysis
- Software-Defined Networks and 5G
- Brain Tumor Detection and Classification
- Advanced Vision and Imaging
- Mobile Ad Hoc Networks
- Wireless Networks and Protocols
- Remote Sensing and LiDAR Applications
- Dielectric materials and actuators
Beihang University
2011-2025
The University of Sydney
2021-2024
Huazhong University of Science and Technology
2022-2024
Children's Medical Research Institute
2021-2024
Ministry of Education of the People's Republic of China
2022-2024
Institute of Advanced Manufacturing Technology
2024
Affiliated Hospital of Qingdao University
2022-2023
Qingdao University
2018-2023
Tongji University
2023
Yanshan University
2023
The rapidly decreasing computation and memory cost has recently driven the success of many applications in field deep learning. Practical learning resource-limited hardware, such as embedded devices smart phones, however, remain challenging. For binary convolutional networks, reason lies degraded representation caused by binarizing full-precision filters. To address this problem, we propose new circulant filters (CiFs) a convolution (CBConv) to enhance capacity binarized features via our...
Multimodal single-cell omics technologies enable multiple molecular programs to be simultaneously profiled at a global scale in individual cells, creating opportunities study biological systems resolution that was previously inaccessible. However, the analysis of multimodal data is challenging due lack methods can integrate across modalities generated from such technologies. Here, we present Matilda, multi-task learning method for integrative data. By leveraging interrelationship among...
Cuproptosis is a newly recognized copper-dependent nonapoptotic form of cell death, which stimulate studies exploring copper-based nanomaterials to treat cancer through distinct mechanistic action. However, it remains challenge completely eradicate tumors via monotherapy. Herein, copper-doped BiSex (CBS) nanozyme was developed boost αPD-L1-mediated immune checkpoint blocking (ICB) synergetic apoptosis/cuproptosis-induced immunogenic death (ICD). The defect-engineered CBS exhibits strong...
This study proposes a novel method for image registration and fusion via commonly used visible light infrared integrated cameras mounted on medium-altitude unmanned aerial vehicles (UAVs).The innovation of lies in three aspects. First, it reveals how complex perspective transformation can be converted to simple scale translation between two sensor images under long-distance parallel imaging conditions. Second, with the introduction metadata, calculation algorithm is designed according...
2,5-Furandicarboxylic acid (FDCA) is a bio-based platform chemical for the production of polyethylene furanoate (PEF) and other valuable furanic chemicals. A magnetic laccase catalyst with (2,2,6,6-tetramethyl-piperidin-1-yl)oxyl (TEMPO) as mediator has remarkable capability oxidizing 5-hydroxymethylfurfural (HMF) to 2,5-furandicarboxylic (FDCA). Under optimal reaction conditions, quantitative yield (90.2 %) FDCA complete HMF conversion was obtained after 96 h reaction. More importantly,...
In this paper, we find that the conventional convolution operation becomes bottleneck for extremely efficient binary neural networks (BNNs). To address issue, open up a new direction by introducing reshaped point-wise (RPC) to replace one build BNNs. Specifically, conduct after rearranging spatial information into depth, with which at least <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$2.25\times $...
Feature selection is an essential task in single-cell RNA-seq (scRNA-seq) data analysis and can be critical for gene dimension reduction downstream analyses, such as marker identification cell type classification. Most popular methods feature from scRNA-seq are based on the concept of differential distribution wherein a statistical model used to detect changes expression among types. Recent development deep learning-based provides alternative approach compared traditional distribution-based...
Objectives This study examined the protective effect of salubrinal and mechanism underlying this protection on tunicamycin (TM)- hypoxia-induced apoptosis in rat cardiomyocytes. Methods Neonatal cardiomyocytes were cultured from ventricles 1-day-old Wistar rats. Cells exposed to different concentrations (10, 20, 40 μmol/L) for 30 minutes followed by TM treatment or hypoxia 36 hours. Apoptosis was measured a multiparameter HCS (high content screening) assay, TUNEL assay flow cytometry. The...
As the third generation of neural networks, Spiking Neural Networks (SNNs) have gained widespread attention due to their low energy consumption and biological interpretability. Recently, SNNs made considerable advancements in computer vision. However, efficiently conducting feature extraction fusion under spiking characteristics for object detection remains a pressing challenge. To address this problem, we propose SpikSSD, novel Single Shot Multibox Detector. Specifically, design...
Advancements in aviation technology have made intelligent navigation systems essential for improving flight safety and efficiency, particularly low-visibility conditions. Radar GPS face limitations bad weather, making visible–infrared sensor fusion a promising alternative. This study proposes salient object detection (SOD) method that integrates visible infrared sensors robust airport runway complex environments. We introduce large-scale dataset (RDD5000) develop SOD algorithm capable of...
Ternary neural networks (TNNs) are potential for network acceleration by reducing the full-precision weights in to ternary ones, e.g., {-1,0,1}. However, existing TNNs mostly calculated based on rule-of-thumb quantization methods simply thresholding operations, which causes a significant accuracy loss. In this paper, we introduce stem-residual framework provides new insight into quantization, termed Residual Quantization (TRQ), achieve more powerful TNNs. Rather than directly TRQ recursively...
Integrating multi-modal data can significantly increase detection performance in a complex scene by introducing additional targets' information. However, most of the existing detectors separately extract features from respective modalities without regarding correlation between modalities. Considering spatial across different for aligned data, we attempt to exploit such share target's information modalities, thereby enhancing feature representation capability. To this end, letter, propose an...
Recent advances in multimodal single-cell omics technologies enable multiple modalities of molecular attributes, such as gene expression, chromatin accessibility, and protein abundance, to be profiled simultaneously at a global level individual cells. While the increasing availability data is expected provide more accurate clustering characterization cells, development computational methods that are capable extracting information embedded across still its infancy.We propose SnapCCESS for...