检索结果-南通市图书馆

基于TransMANet的遥感图像语义分割算法

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

激光与光电子学进展 2024年第10期61卷 301-312页

作者：宋熙睿葛洪伟江南大学人工智能与计算机学院江苏无锡214122 江苏省模式识别与计算智能工程实验室(江南大学) 江苏无锡214122

针对multiattention network(MANet)算法与图像语义信息关联不足、全局特征提取不充分和分割精度较低的问题,基于Transformer与注意力机制,提出一种增强浅层网络语义信息,具有融合局部和全局上下文的双分支解码器的网络结构,即Transform... 详细信息

针对multiattention network(MANet)算法与图像语义信息关联不足、全局特征提取不充分和分割精度较低的问题,基于Transformer与注意力机制,提出一种增强浅层网络语义信息,具有融合局部和全局上下文的双分支解码器的网络结构,即Transformer multiattention network(TransMANet)。首先,引入局部注意力嵌入机制,增强上下文信息的嵌入,并将高级特征的语义信息嵌入低级特征;然后,设计基于Transformer与卷积神经网络的双分支解码器,分别提取全局上下文信息和不同尺度的细节信息,对全局与局部信息建模;最后,改进原有的损失函数,缓解遥感数据集类别不平衡的问题,提高分割准确度。实验结果表明,TransMANet在UAVid、LoveDA、Potsdam和Vaihingen数据集上均取得了较MANet及其他有竞争力的先进方法更优的交并比指标,有较好的泛化能力。

关键词：图像处理语义分割注意力机制 Transformer 高分辨率遥感影像

在线全文

学校读者我要写书评

暂无评论

纵横比自适应的时空正则化相关滤波算法

Journal of Measurement Science and Instrumentation 2024年第1期15卷 9-22页

作者：许凯李婷葛洪伟江南大学人工智能与计算机学院江苏无锡214122 江苏省模式识别与计算智能工程实验室(江南大学) 江苏无锡214122

在目标跟踪中,传统相关滤波算法无法感知运动目标尺度纵横比变化,且易受复杂环境影响导致跟踪失败。为此,提出了纵横比自适应的时空正则化相关滤波算法。首先,参考平均峰值相关能量(Average peak-to-correlation energy,APCE)和响应峰... 详细信息

在目标跟踪中,传统相关滤波算法无法感知运动目标尺度纵横比变化,且易受复杂环境影响导致跟踪失败。为此,提出了纵横比自适应的时空正则化相关滤波算法。首先,参考平均峰值相关能量(Average peak-to-correlation energy,APCE)和响应峰值对每个特征的响应图进行加权融合,以实现对目标的精确跟踪。其次,结合近正交性和空间正则化提出一种新的一维边界滤波器,通过定位目标包围框的四个边界位置实现对目标尺度和纵横比变化的自适应检测,有效抑制了边界效应带来的负面影响。最后,根据响应输出的峰值旁瓣比(Peak-to-sidelobe ratio,PSR)独立地调节各边界滤波器的学习率,防止模型退化。在OTB数据集上进行了测试,该算法表现出理想的跟踪效果,在各个挑战属性上相较于其他优秀算法均取得了更优结果。

关键词：目标跟踪相关滤波空间正则化自适应纵横比模板更新响应融合

在线全文

学校读者我要写书评

暂无评论

基于EMA改进的图像语义分割算法

Journal of Measurement Science and Instrumentation 2024年第2期15卷 185-194页

作者：杜佳栋李婷葛洪伟江南大学人工智能与计算机学院江苏无锡214122 江南大学江苏省模式识别与计算智能工程实验室江苏无锡214122

针对期望最大化注意(EMA)算法参数与图像的语义关联不足以及缺少对通道间信息关注的问题,本文提出一种双重注意力网络EMA+算法。该算法设计了2个模块:空间注意力模块和通道注意力模块。空间注意力模块以EMA算法为主体架构,在责任估计步... 详细信息

针对期望最大化注意(EMA)算法参数与图像的语义关联不足以及缺少对通道间信息关注的问题,本文提出一种双重注意力网络EMA+算法。该算法设计了2个模块:空间注意力模块和通道注意力模块。空间注意力模块以EMA算法为主体架构,在责任估计步骤采用特征图作为期望最大化(EM)算法的初始参数,增加参数与特征图语义上的关联。通道注意力模块使用高效通道注意力(ECA),通过使用一维卷积学习通道之间交互信息,避免由于降维操作导致的破坏通道与其权重之间的直接对应关系。EMA+通过融合空间注意力模块和通道注意力模块,显著提高了语义分割任务的性能。实验结果表明,EMA+在PASCAL VOC2012和一些更复杂的数据集上均取得了较EMANet等方法更优的交并比指标,有较好的泛化能力。

关键词：深度学习图像语义分割期望最大化注意双重注意力网络高效通道注意力模块

在线全文

学校读者我要写书评

暂无评论

经方治疗颈椎病思路探析

中华中医药杂志 2024年第1期39卷 268-271页

作者：黄悠葛宏伟何晓瑾南京中医药大学附属医院江苏省中医院南京210029

颈椎病是临床上一种常见病和多发病,属于中医学“痹证”“头痛”“眩晕”等范畴。对于颈椎病的治疗,经方有着独特的优势,且从古至今已积累了丰富的治疗经验。文章从经方的方证视角出发,通过遴选《伤寒论》和《金匮要略》条文中出现的与... 详细信息

颈椎病是临床上一种常见病和多发病,属于中医学“痹证”“头痛”“眩晕”等范畴。对于颈椎病的治疗,经方有着独特的优势,且从古至今已积累了丰富的治疗经验。文章从经方的方证视角出发,通过遴选《伤寒论》和《金匮要略》条文中出现的与颈椎病症状相关的方剂,深入探讨其病机、经络及组方思路,总结出桂枝-麻黄类方、葛根类方、柴胡类方、附子类方、苓桂类方等几大类方,探析如何从经方入手治疗颈椎病。

关键词：颈椎病经方方证痹证眩晕

Significantly enhanced hydrogen desorption properties of Mg（AlH4）2 nanoparticles synthesized using solvent free strategy

在线全文

学校读者我要写书评

暂无评论

Progress in Natural Science:Materials International 2017年第1期27卷 112-120页

作者： Xuezhang Xiao Teng Qin Yiqun Jiang Fulei Jiang Meng Li Xiulin Fan Shouquan Li hongwei ge Qidong Wang Lixin Chen State Key Laboratory of Silicon Materials Key Laboratory of Advanced Materials and Applications for Batteries of Zhejiang Province School of Materials Science and Engineering Zhejiang University

Mg（AlH4）2nanoparticles with a particle size less than 10 nm have been successfully synthesized by mechanochemical method using LiAlH4and MgCl2as raw materials together with Li Cl buffering additive. In comparison to Mg（AlH4）2microparticles, Mg（AlH4）2nanoparticles exhibit a faster hydrogen desorption kinetics and lower desorption temperature. The hydrogen desorption temperatures of the first and second dehydrogenation steps are 80 and 220 °C for the Mg（AlH4）2nanoparticles, which are about 65 and 60 °C, respectively, lower than those of Mg（AlH4）2microparticles. The decomposition activation energy is reduced from 135 k J/mol for Mg（AlH4）2microparticles to 105.3 k J/mol for Mg（AlH4）2nanoparticles. It is proposed that the shortened diffusion distance and enhanced diffusivity of Mg（AlH4）2/MgH2nanoparticles provide an energy destabilization for lowering the dehydrogenation temperature, and thus being the key factor for promoting the hydrogen desorption kinetics. More importantly, it is demonstrated that the dehydrided nano MgH2hydride with a particle size below 10 nm can be formed after rehydrogenation process, resulting in the good cycling hydrogen desorption performance of nano MgH2.

关键词： Hydrogen storage Complex hydride Mg(AlH4)2 Nanoparticle Kinetics

Unpaired image to image transformation via informative coupled generative adversarial networks

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2021年第4期15卷 83-92页

作者： hongwei ge Yuxuan HAN Wenjing KANG Liang SUN College of Computer Science and Technology Dalian University of TechnologyDalian 116024China

We consider image transformation problems,and the objective is to translate images from a source domain to a target one.The problem is challenging since it is difficult to preserve the key properties of the source images,and to make the details of target being as distinguishable as possible.To solve this problem,we propose an informative coupled generative adversarial networks(ICoGAN).For each domain,an adversarial generator-and-discriminator network is constructed.Basically,we make an approximately-shared latent space assumption by a mutual information mechanism,which enables the algorithm to learn representations of both domains in unsupervised setting,and to transform the key properties of images from source to target.Moreover,to further enhance the performance,a weightsharing constraint between two subnetworks,and different level perceptual losses extracted from the intermediate layers of the networks are combined.With quantitative and visual results presented on the tasks of edge to photo transformation,face attribute transfer,and image inpainting,we demonstrate the ICo-GAN’s effectiveness,as compared with other state-of-the-art algorithms.

关键词： generative adversarial networks image transformation mutual information perceptual loss

A Quantum Spatial Graph Convolutional Network for Text Classification

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computer Systems Science & Engineering 2021年第2期36卷 369-382页

作者： Syed Mustajar Ahmad Shah hongwei ge Sami Ahmed Haider Muhammad Irshad Sohail M.Noman Jehangir Arshad Asfandeyar Ahmad Talha Younas College of Computer Science and Technology Dalian University of TechnologyDalian116024China Department of Computing University of WorcesterUK Department of Information Engineering Chang’an UniversityXi’an710054China Department of Cell Biology and genetics Shantou University Medical CollegeShantou515041China Electrical and Computer Engineering Department COMSATS University IslamabadLahore CampusPakistan School of Information&Communication Engineering Dalian University of Technology Dalian116024China Electrical and Computer Engineering Department COMSATS University IslamabadSahiwal CampusPakistan

The data generated from non-Euclidean domains and its graphical representation(with complex-relationship object interdependence)applications has observed an exponential growth.The sophistication of graph data has posed consequential obstacles to the existing machine learning algorithms.In this study,we have considered a revamped version of a semi-supervised learning algorithm for graph-structured data to address the issue of expanding deep learning approaches to represent the graph data.Additionally,the quantum information theory has been applied through Graph Neural Networks(GNNs)to generate Riemannian metrics in closed-form of several graph layers.In further,to pre-process the adjacency matrix of graphs,a new formulation is established to incorporate high order proximities.The proposed scheme has shown outstanding improvements to overcome the deficiencies in Graph Convolutional Network(GCN),particularly,the information loss and imprecise information representation with acceptable computational overhead.Moreover,the proposed Quantum Graph Convolutional Network(QGCN)has significantly strengthened the GCN on semi-supervised node classification tasks.In parallel,it expands the generalization process with a significant difference by making small random perturbationsG of the graph during the training process.The evaluation results are provided on three benchmark datasets,including Citeseer,Cora,and PubMed,that distinctly delineate the superiority of the proposed model in terms of computational accuracy against state-of-the-art GCN and three other methods based on the same algorithms in the existing literature.

关键词： Text classiﬁcation deep learning graph convolutional networks semi-supervised learning GPUs performance improvements

Atomic-level unveiling secondary recrystallization enabled microand macroscopic polarization enhancement for piezophotocatalytic oxygen activation

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Nano Research 2024年第6期17卷 5040-5049页

作者： Kai Lin Zijian Zhu Weiyi ge Tianxing Jiang hongwei Huang Engineering Research Center of Ministry of Education for geological Carbon Storage and Low Carbon Utilization of Resources Beijing Key Laboratory of Materials Utilization of Nonmetallic Minerals and Solid WastesNational Laboratory of Mineral MaterialsSchool of Material Sciences and TechnologyChina University of Geosciences(Beijing)Beijing 100083China School of Chemistry Beihang UniversityBeijing 100191China

Piezoelectric semiconductors bear the bifunctional photocatalysis and piezocatalysis,while the absent or weak internal charge driving force severely restricts its catalytic activity.Developing polarization strategy is desirable,and particularly understanding its mechanism from a microscopic perspective remains scanty.Herein,we report a secondary recrystallization approach to achieving the simultaneous micro-and macroscopic polarization enhancement on Bi2WO6 nanosheets for boosting piezo-photocatalytic oxygen activation,and unravel the mechanism at an atom-level.The secondary recrystallization process not only results in a strengthened distortion of[WO6]octahedra with distortion index enhancement by~20%for a single octahedron,but also enables lateral crystal growth of nanosheets along the ab plane(av.50 to 180 nm),which separately allows the rise in dipole moment of unit cell(e.g.,1.63 D increase along a axis)and the stacking of the distorted[WO6]octahedron to accumulate the unit cell dipole,collectively contributing to the considerably strengthened spontaneous polarization and piezoelectricity.Besides,exposure of large-area{001}front facet enables more efficient capture and conversion of stress into piezo-potential.Therefore,the well-recrystallized Bi2WO6 nanosheets exhibit considerably promoted piezo-photocatalytic reactive oxygen species generation,given the decreased specific surface area.This work presents a feasible methodology to regulate inside-out polarization for guiding carriers transfer behavior,and may advance the solid understanding on the intrinsic mechanism.

关键词： secondary recrystallization piezo-photocatalytic oxygen activation octahedron distortion polar units stacking micro-and macroscopic polarization enhancement

标本溶血对血站血液筛查实验室ELISA检测结果的影响

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

中国输血杂志 2024年第4期37卷 417-424页

作者：秦倩倩葛红卫张婧赵子复冷婵王瑞北京市红十字血液中心北京100088

目的分析标本溶血对血站实验室ELISA检测结果的影响,确定针对实验室试验项目及检测系统特异的可接受的溶血程度,为制定溶血标本接受标准提供依据。方法针对HBsAg、HIV Ag/Ab、抗-HCV、抗-TP 4项ELISA试验,分别制备不同溶血梯度的阴性... 详细信息

目的分析标本溶血对血站实验室ELISA检测结果的影响,确定针对实验室试验项目及检测系统特异的可接受的溶血程度,为制定溶血标本接受标准提供依据。方法针对HBsAg、HIV Ag/Ab、抗-HCV、抗-TP 4项ELISA试验,分别制备不同溶血梯度的阴性标本和弱阳性标本(S/CO≈2),使用几种常用的国产试剂对上述标本进行检测,分析不同程度溶血对各项目阴性和弱阳性标本检测结果的影响。结果 1)标本溶血对HBsAg、抗-HCV、抗-TP ELISA试验阴性和弱阳性标本的检测结果(反应性/非反应性)无影响;2)标本溶血对HIV Ag/Ab ELISA试验阴性和弱阳性标本的检测结果(反应性/非反应性)存在影响,建议将Hb 2 g/L作为HIV Ag/Ab ELISA试验可接受的溶血程度。结论本研究确定了适用于本实验室相应试验项目和检测系统的溶血标本接受标准。溶血对血站实验室ELISA检测结果的影响与试剂、设备、环境等多种因素有关,实验室应结合自身具体情况进行评估和验证,科学合理地确定溶血标本接受标准。

关键词：溶血血站实验室 ELISA