检索结果-内蒙古大学图书馆

Knowledge distillation of face recognition via attention cosine similarity review

IET COMPUTER VISION 2024年第7期18卷 875-887页

作者： Wang, Zhuo zhao, suwen Guo, Wanyi Xian Univ Architecture & Technol Coll Informat & Control Engn Xian Peoples R China

Deep learning-based face recognition models have demonstrated remarkable performance in benchmark tests, and knowledge distillation technology has been frequently accustomed to obtain high-precision real-time face recognition models specifically designed for mobile and embedded devices. However, in recent years, the knowledge distillation methods for face recognition, which mainly focus on feature or logit knowledge distillation techniques, neglect the attention mechanism that play an important role in the domain of neural networks. An innovation cross-stage connection review path of the attention cosine similarity knowledge distillation method that unites the attention mechanism with review knowledge distillation method is proposed. This method transfers the attention map obtained from the teacher network to the student through a cross-stage connection path. The efficacy and excellence of the proposed algorithm are demonstrated in popular benchmark tests. The authors propose a cross-stage connection review of the attention cosine similarity knowledge distillation method. This method leverages the teacher's shallow multi-stage attention maps to guide deep single-stage attention map in the student model. In the AMFM module, multi-stage attention map is fused, and in the HCCSL module, the spatial pyramid pooling technique is employed obtain multi-level of attention map information, different levels of attention context information and attention map are calculated cosine similarity loss. The proposed algorithm superior performance is compared to sota methods. image

关键词： computer vision convolutional neural nets face recognition

来源：评论

学校读者我要写书评

暂无评论

Target tracking algorithm based on multiscale analysis and combinatorial matching

引用

JOURNAL OF SUPERCOMPUTING 2022年第10期78卷 12648-12661页

作者： Wang, Yanni Guo, Rongchun zhao, suwen Xian Univ Architecture & Technol Sch Informat & Control Engn Xian 710055 Shaanxi Peoples R China Xian Univ Posts & Telecommun Sch Digital Arts Xian 710121 Shaanxi Peoples R China

Aiming at improving the accuracy of current tracking algorithms and tracking the target more quickly, a new target tracking method that is from multiscale analysis is presented, which is combined with combinatorial polygon matching. First, a target's range is estimated by analyzing multiple scales in the reference frame. Then, the pixels are judged as the target pixels within the target range determined in the previous step by calculating the relevant radial distance and judging the similarity of radial distance between target pixels to be detected and its background pixels. At the same time, the background modeling is updated according to the similarity between the incremental code of the background pixels and the corresponding pixels in the observed images. Last, the target trajectory is obtained by dynamic polygon matching. When encountering external interference, the performance of the improved algorithm is relatively stable. Compared with the existing methods, the proposed algorithm in this paper is verified by simulation, which is simple in operation and can track the target quickly and accurately.

关键词： Target tracking Multiscale analysis Combinatorial polygon matching Relevant radial distance

来源：评论

学校读者我要写书评

暂无评论

Peptide-Scaffolded Detergents for Membrane Protein Studies

引用

CHEMISTRY-A EUROPEAN JOURNAL 2025年第16期31卷 e202404520页

作者： Yang, Meifang Dai, Yili Zhou, Fang Zhou, Xin Qiu, Yanli Tan, Yan zhao, suwen Xue, Dongxiang zhao, Fei Tao, Houchao Shanghai Univ Tradit Chinese Med Innovat Res Inst Tradit Chinese Med Shanghai Frontiers Sci Ctr TCM Chem Biol Shanghai 201203 Peoples R China Univ South China Inst Pharm & Pharmacol Hengyang Med Sch Hengyang 421001 Hunan Peoples R China ShanghaiTech Univ iHuman Inst Shanghai 201210 Peoples R China

Detergents are essential for preserving the structural integrity and functionality of membrane proteins (MPs) outside the biological membrane or in aqueous solution, and thus ensuring accurate biochemical and structural analyses. Here, we introduce peptide-scaffolded detergents, a novel class of hybrid molecules formed by preassembling detergent monomers with peptides of varying lengths, mediated via Click chemistry. These detergents are characterized by scalable, straightforward synthesis and enhanced solubility. Among the variants, A4B2 emerged as the optimal detergent, demonstrating superior thermal stabilization across a range of G protein-coupled receptors, including A2AAR, SMO and GLP-1R. Additionally, A4B2 exhibits a low critical micelle concentration and small micelle size, together making it particularly effective for electron microscopy studies of A2AAR. This innovative design leverages the benefits of peptide-based and traditional detergents, offering new insights for the development of advanced detergents in MP research.

关键词： Peptide-scaffolded detergents Membrane proteins Micelle Click reaction

来源：评论

学校读者我要写书评

暂无评论

Building the Chordata Olfactory Receptor Database using more than 400,000 receptors annotated by Genome2OR

引用

Science China(Life Sciences) 2022年第12期65卷 2539-2551页

作者： Wei Han Yiran Wu Liting Zeng suwen zhao iHuman Institute ShanghaiTech UniversityShanghai 201210China School of Life Science and Technology ShanghaiTech UniversityShanghai 201210China University of Chinese Academy of Sciences Beijing 100049China Shanghai Institute of Nutrition and Health Chinese Academy of SciencesShanghai 200031China

Olfactory receptors are poorly annotated for most genome-sequenced *** address this deficiency,we developed a nhmmer-based olfactory receptor annotation tool Genome2OR(https://***/To Hanwei/***),and used it to process 1,695 sequenced chordate genomes in the NCBI Assembly database as of January,*** total,765,248 olfactory receptor genes were annotated,with 404,426 functional genes and 360,822 pseudogenes,which represents a four-fold increase in the number of annotated olfactory *** on the annotation data,we built a database called Chordata Olfactory Receptor Database(CORD,https://***)for archiving,analysing and disseminating the *** the primary data,we offer derivative information,including pictures of species,cross references to public databases,structural models,sequence similarity networks and sequence profiles in the ***,we did brief analyses on these receptors,including building a huge protein sequence similarity network covering all receptors in the database,and clustering them into 20 communities,classifying the 20 communities into three categories based on their presences/absences in ray-finned fish and/or lobe-finned *** infer that olfactory receptors should have unique activation and desensitization mechanisms by analysing their sequences and structural *** believe the CORD can benefit the researchers and the general public who are interested in olfaction.

关键词： G protein-coupled receptors olfactory receptors Genome2OR CORD

来源：评论

学校读者我要写书评

暂无评论

Large-Scale Analysis of Bioactive Ligand Conformational Strain Energy by Ab Initio Calculation

引用

JOURNAL OF CHEMICAL INFORMATION AND MODELING 2021年第3期61卷 1180-1192页

作者： Tong, Jiahui zhao, suwen ShanghaiTech Univ iHuman Inst Shanghai 201210 Peoples R China ShanghaiTech Univ Sch Life Sci & Technol Shanghai 201210 Peoples R China Univ Chinese Acad Sci Beijing 100049 Peoples R China Chinese Acad Sci Shanghai Inst Nutr & Hlth Shanghai 200031 Peoples R China

Ligand conformational strain energy (LCSE) plays an important role in virtual screening and lead optimization. While various studies have provided insights into LCSE for small-molecule ligands in the Protein Data Bank (PDB), conclusions are inconsistent mainly due to small datasets, poor quality control of crystal structures, and molecular mechanics (MM) or low-level quantum mechanics (QM) calculations. Here, we built a high-quality dataset (LigBoundConf) of 8145 ligand-bound conformations from PDB crystal structures and calculated LCSE at the M062X-D3/ma-TZVPP (SMD)//M062X-D3/def2-SVP(SMD) level for each case in the dataset. The mean/median LCSE is 4.6/3.7 kcal/mol for 6672 successfully calculated cases, which is significantly lower than the estimates based on molecular mechanics in many previous analyses. Especially, when removing ligands with nonaromatic ring(s) that are prone to have large LCSEs due to electron density overfitting, the mean/median LCSE was reduced to 3.3/2.5 kcal/mol. We further reveal that LCSE is correlated with several ligand properties, including formal atomic charge, molecular weight, number of rotatable bonds, and number of hydrogen-bond donors and acceptors. In addition, our results show that although summation of torsion strains is a good approximation of LCSE for most cases, for a small fraction (about 6%) of our dataset, it underestimates LCSEs if ligands could form nonlocal intramolecular interactions in the unbound state. Taken together, our work provides a comprehensive profile of LCSE for ligands in PDB, which could help ligand conformation generation, ligand docking pose evaluation, and lead optimization.

关键词： Ligands Energy Conformation Molecular structure Electron density

来源：评论

学校读者我要写书评

暂无评论

A semi-supervised self-training method based on density peaks and natural neighbors

引用

JOURNAL OF AMBIENT INTELLIGENCE AND HUMANIZED COMPUTING 2021年第2期12卷 2939-2953页

作者： zhao, suwen Li, Junnan Chongqing Univ Coll Bioengn Chongqing 400044 Peoples R China Guilin Univ Aerosp Technol Dept Elect Engn Guilin 541004 Peoples R China Chongqing Univ Coll Comp Sci Chongqing Key Lab Software Theory & Technol Chongqing 400044 Peoples R China

The semi-supervised self-training method is one of the successful methodologies of semi-supervised classification and can train a classifier by exploiting both labeled data and unlabeled data. However, most of the self-training methods are limited by the distribution of initial labeled data, heavily rely on parameters and have the poor ability of prediction in the self-training process. To solve these problems, a novel self-training method based on density peaks and natural neighbors (STDPNaN) is proposed. In STDPNaN, an improved parameter-free density peaks clustering (DPCNaN) is firstly presented by introducing natural neighbors. The DPCNaN can reveal the real structure and distribution of data without any parameter, and then helps STDPNaN restore the real data space with the spherical or non-spherical distribution. Also, an ensemble classifier is employed to improve the predictive ability of STDPNaN in the self-training process. Intensive experiments show that (a) STDPNaN outperforms state-of-the-art methods in improving classification accuracy ofknearest neighbor, support vector machine and classification and regression tree;(b) STDPNaN also outperforms comparison methods without any restriction on the number of labeled data;(c) the running time of STDPNaN is acceptable.

关键词： Self-training method Semi-supervised classification Semi-supervised learning Natural neighbors Density peaks

来源：评论

学校读者我要写书评

暂无评论

Targeting collagen damage for sustained in situ antimicrobial activities

引用

JOURNAL OF CONTROLLED RELEASE 2023年第1期360卷 122-132页

作者： Mo, Xiaoyun zhao, suwen zhao, Jie Huang, Yongjie Li, Tao Zhu, Yongqiao Li, Gang Li, Yang Shan, Hong Sun Yat Sen Univ Guangdong Prov Engn Res Ctr Mol Imaging Cardiac Surg & Struct Heart Dis Unit Cardiovasc Ctr Zhuhai 519000 Peoples R China Sun Yat Sen Univ Affiliated Hosp 5 Dept Radiol Zhuhai 519000 Peoples R China Sun Yat Sen Univ Affiliated Hosp 5 Dept Intervent Med Zhuhai 519000 Peoples R China

Antimicrobial peptides (AMPs) are promising anti-infective drugs, but their use is restricted by their short-term retention at the infection site, non-targeted uptake, and adverse effects on normal tissues. Since infection often follows an injury (e.g., in a wound bed), directly immobilizing AMPs to the damaged collagenous matrix of the injured tissues may help overcome these limitations by transforming the extracellular matrix microenvironment of the infection site into a natural reservoir of AMPs for sustained in situ release. Here, we developed and demonstrated an AMP-delivery strategy by conjugating a dimeric construct of AMP Feleucin-K3 (Flc) and a collagen hybridizing peptide (CHP), which enabled selective and prolonged anchoring of the Flc-CHP conjugate to the damaged and denatured collagen in the infected wounds in vitro and in vivo. We found that the dimeric Flc and CHP conjugate design preserved the potent and broad-spectrum antimicrobial activities of Flc while significantly enhancing and extending its antimicrobial efficacy in vivo and facilitating tissue repair in a rat wound healing model. Because collagen damage is ubiquitous in almost all injuries and infections, our strategy of targeting collagen damage may open up new avenues for antimicrobial treatments in a range of infected tissues.

关键词： Peptide bioconjugate Antimicrobial peptide Collagen hybridizing peptide Triple-helix Wound healing Denatured collagen

来源：评论

学校读者我要写书评

暂无评论

Controlling the Trimerization of the Collagen Triple-Helix by Solvent Switching

引用

BIOMACROMOLECULES 2023年第4期24卷 1689-1699页

作者： Zhang, Qi Li, Xiaojing Huang, Kui Huang, Yongjie zhao, suwen Liu, Shanshan Li, Yang Sun Yat Sen Univ Affiliated Hosp 5 Guangdong Prov Engn Res Ctr Mol Imaging Zhuhai 519000 Peoples R China Sun Yat Sen Univ Affiliated Hosp 5 Dept Crit Care Med Surg Intens Care Unit Zhuhai 519000 Peoples R China Sun Yat Sen Univ Affiliated Hosp 5 Dept Radiol Zhuhai 519000 Peoples R China Sun Yat Sen Univ Affiliated Hosp 5 Cardiovasc Ctr Cardiac Surg & Struct Heart Dis Unit Zhuhai 519000 Peoples R China

Collagen hybridizing peptides (CHPs) are a powerful tool for targeting collagen damage in pathological tissues due to their ability to specifically form a hybrid collagen triple-helix with the denatured collagen chains. However, CHPs have a strong tendency to self-trimerize, requiring preheating or complicated chemical modifications to dissociate their homotrimers into monomers, which hinders their applications. To control the self assembly of CHP monomers, we evaluated the effects of 22 cosolvents on the triple-helix structure: unlike typical globular proteins, the CHP homotrimers (as well as the hybrid CHP- collagen triple helix) cannot be destabilized by the hydrophobic alcohols and detergents (e.g., SDS) but can be effectively dissociated by the cosolvents that dominate hydrogen bonds (e.g., urea, guanidinium salts, and hexafluoroisopropanol). Our study provided a reference for the solvent effects on natural collagen and a simple effective solvent-switch method, enabling CHP utilization in automated histopathology staining and in vivo imaging and targeting of collagen damage.

关键词： Animal derived food Biopolymers Peptides and proteins Solvents Urea

来源：评论

学校读者我要写书评

暂无评论

"Living Detergents": an in Situ Detergent Tailoring Strategy for Efficient Membrane Protein Stabilization and Analysis

引用

CHEMISTRY-A EUROPEAN JOURNAL 2025年第28期31卷 e202501128页

作者： Dai, Yili Yang, Meifang Luo, Weiling Qiu, Yanli Zhou, Fang Zheng, Xing zhao, Fei Yao, Xu zhao, suwen Tao, Houchao Univ South China Inst Pharm & Pharmacol Hengyang Med Sch Hengyang 421001 Hunan Peoples R China Shanghai Univ Tradit Chinese Med Innovat Res Inst Tradit Chinese Med Shanghai Frontiers Sci Ctr TCM Chem Biol Shanghai 201203 Peoples R China ShanghaiTech Univ iHuman Inst Shanghai 201210 Peoples R China Hunan Vocat Coll Sci & Technol Dept Pharm Changsha 410004 Hunan Peoples R China

Detergents are essential molecular tools for membrane protein (MP) research, yet traditional detergents with static properties often fail to address the diverse and evolving needs of MP studies. To this end, this study introduces "living detergents", an innovative class of detergents equipped with functional tags that enable bioorthogonal modifications with externally introduced structural elements. This approach allows for not only the parallel generation of new detergents, but also in situ tuning of MP samples within freshly formed detergents. The efficacy of this strategy was demonstrated through the rapid identification of optimal detergents for high-quality electron microscopy studies of A2AAR. Overall, this flexible and robust platform enables efficient tailoring of detergents, advancing the exploration of detergent structure-function relationships in MP research and opening pathways for more specialized solutions for diverse experimental demands.

关键词： bioorthogonal reaction living detergent membrane proteins micelle Introduction

来源：评论

学校读者我要写书评

暂无评论

Structural and biofunctional evaluation of decellularized jellyfish matrices

引用

JOURNAL OF MATERIALS CHEMISTRY B 2023年第16期11卷 3740-3751页

作者： zhao, Jie Yu, Wenjun Zhang, Qi Li, Xiaojing Huang, Yongjie zhao, suwen Li, Tao Liu, Shanshan Li, Yang Shan, Hong Sun Yat Sen Univ Affiliated Hosp 5 Guangdong Prov Engn Res Ctr Mol Imaging Zhuhai 519000 Peoples R China Sun Yat Sen Univ Affiliated Hosp 5 Dept Radiol Zhuhai 519000 Peoples R China Sun Yat Sen Univ Affiliated Hosp 5 Cardiovasc Ctr Cardiac Surg & Struct Heart Dis Unit Zhuhai 519000 Peoples R China Sun Yat Sen Univ Affiliated Hosp 5 Dept Crit Care Med Surg Intens Care Unit Zhuhai 519000 Peoples R China Sun Yat Sen Univ Affiliated Hosp 5 Dept Intervent Med Zhuhai 519000 Peoples R China

Extracellular matrices decellularized from marine animal tissues are emerging scaffolds in tissue engineering. Jellyfish tissues are suitable for making functional and safe decellularized matrices in part due to their simple structure, high water content, and low risk of pathogen transmission to humans. Jellyfish are some of the most prevalent marine animals, but their decellularized matrices have remained largely undeveloped. Here we evaluated the structures and functions of the jellyfish (Rhopilema esculentum) matrices decellularized with seven different detergents. All of them showed effectiveness in removing the cellular components. Scanning electron microscopy and mechanical testing revealed that the decellularized matrices mostly retained the native microstructures, whereas only SDS and SNL distorted the matrices' multilayered and fibrous architecture. The collagen hybridizing peptide fluorescence staining showed that SDS, SNL, Triton X-100, IGEPAL, and Tween-20 denatured the jellyfish collagen molecules to varying degrees while CHAPS and SD protected the collagen's triple-helix conformation. We demonstrated that the decellularized jellyfish matrices showed similarity to different types of mammalian collagen and supported the adhesion and proliferation of human dermal and corneal fibroblasts and mouse chondrocytes in 3D culture. Importantly, the decellularized jellyfish matrix also facilitated wound healing in vivo by reducing inflammation while promoting angiogenesis and tissue remodeling. Taken together, our study demonstrated that the decellularized jellyfish matrices are an easy-to-prepare, biocompatible, and potentially widely applicable scaffold for regenerative medicine.

关键词： Collagen

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：