检索结果-南通市图书馆

Design and implementation of dual-mode configurable memory architecture for CNN accelerator

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

High Technology Letters 2024年第2期30卷 211-220页

作者：山蕊 LI Xiaoshuo GAO Xu huo ziqing School of Electronic Engineering Xi'an University of Posts and TelecommunicationsXi’an 710121P.R.China

With the rapid development of deep learning algorithms,the computational complexity and functional diversity are increasing ***,the gap between high computational density and insufficient memory bandwidth under the traditional von Neumann architecture is getting *** the algorithmic characteristics of convolutional neural network(CNN),it is found that the access characteristics of convolution(CONV)and fully connected(FC)operations are very *** on this feature,a dual-mode reronfigurable distributed memory architecture for CNN accelerator is *** can be configured in Bank mode or first input first output(FIFO)mode to accommodate the access needs of different *** the same time,a programmable memory control unit is designed,which can effectively control the dual-mode configurable distributed memory architecture by using customized special accessing instructions and reduce the data accessing *** proposed architecture is verified and tested by parallel implementation of some CNN *** experimental results show that the peak bandwidth can reach 13.44 GB·s^(-1)at an operating frequency of 120 *** work can achieve 1.40,1.12,2.80 and 4.70 times the peak bandwidth compared with the existing work.

关键词： distributed memory structure neural network accelerator reconfigurable arrayprocessor configurable memory structure

基于近存储计算的手写数字识别实时检测阵列结构设计

在线全文

学校读者我要写书评

暂无评论

光电子．激光 2022年第12期33卷 1315-1322页

作者：霍紫晴山蕊冯雅妮高旭冯煜西安邮电大学电子工程学院陕西西安710121

卷积神经网络(convolutional neural network, CNN)作为传统神经网络的改进,已经得到了广泛的应用。然而,在CNN性能提升的同时其模型的规模不断扩大,对存储及算力的要求越来越高,基于冯·诺依曼体系结构的处理器难以达到令人满意的高处... 详细信息

卷积神经网络(convolutional neural network, CNN)作为传统神经网络的改进,已经得到了广泛的应用。然而,在CNN性能提升的同时其模型的规模不断扩大,对存储及算力的要求越来越高,基于冯·诺依曼体系结构的处理器难以达到令人满意的高处理性能。为了提升系统性能,近存储计算(near memory computing, NMC)成为了一个具有发展前景的研究方向。本文利用一种支持NMC的可重构阵列处理器实现手写数字识别,并行地实现了卷积运算;同时利用共享缓存阵列结构,减少片外存储的频繁访问。实验结果表明,在110 MHz的工作频率下,执行单个5×5卷积运算的计算速度提升了75.00%,可以在9 960μs内实现一个手写数字的识别。

关键词：卷积神经网络(CNN) 手写数字识别可重构阵列处理器近存储计算(NMC) 共享缓存阵列

Composite Eu-MOF@CQDs“off&on”ratiometric luminescent probe for highly sensitive chiral detection of l-lysine and 2-methoxybenzaldehyde

在线全文

学校读者我要写书评

暂无评论

Chinese Chemical Letters 2023年第11期34卷 231-237页

作者： Yupeng Jiang Xinhui Fang ziqing Zhang Xiaomeng Guo Jianzhong huo Qian Wang Yuanyuan Liu Xinrui Wang Bin Ding Tianjin Key Laboratory of Structure and Performance for Functional Molecule College of ChemistryTianjin Normal UniversityTianjin 300387China

The high amount of l-lysine can increase the potential risk of cardiovascular ***,2-methoxy benzaldehyde(2-MB)has high toxicity and can easily pollute the *** this work,carbon quantum dots(CQDs)can be encapsulated into Eu-BTB(H_(3)BTB=1,3,5-tri(4-carboxyphenyl)benzene),forming the multi-emission composite material Eu-BTB@*** has two emissions peaks(617 nm for Eu and 470 nm for CQDs).Eu-BTB@CQDs can be applied as bi-functional ratiometric“off&on”luminescent sensor for l-lysine and 2-MB with high sensitivity and selectivity,the low limit of detection(LOD)for l-lysine is 3.68μmol/L and for 2-MB is 0.54μmol/L,***,Eu-BTB@CQDs can quantitatively discriminate l-lysine in the mixed d-and l-lysine water solutions(five different concentrations ratio of l/d-lysine has been set)makes the chiral detection of l-lysine are more *** the other hand,Eu-BTB@CQDs also can detect 2-MB over 4-methoxybenzaldehyde(4-MB)with high *** the detection of 2-MB and l-lysine in the lake water real samples with the reasonable recovery ***,the detection mechanisms for l-lysine and 2-MB were also investigated and discussed in detail.

关键词： Photo-luminescence Carbon quantum dots Composite material l-Lysine 2-Methoxybenzaldehyde Isomers

Brain circuits for retching-like behavior

在线全文

学校读者我要写书评

暂无评论

National Science Review 2024年第1期11卷 145-160页

作者： Lifang huo Zhimin Ye Meiling Liu ziqing He Meizhu Huang Dapeng Li Qian Wu Qian Wang Xiaoqun Wang Peng Cao Ji Dong Congping Shang School of Basic Medical Sciences Guangzhou National Laboratory The Fifth Afflliated Hospital Guangzhou Medical University Bioland Laboratory (Guangzhou Regenerative Medicine and Health Guangdong Laboratory) Department of Neurobiology School of Basic Medical Sciences Beijing Key Laboratory of Neural Regeneration and Repair Advanced Innovation Center for Human Brain Protection Capital Medical University State Key Laboratory of Cognitive Neuroscience and Learning IDG/McGovern Institute for Brain Research Beijing Normal University Changping Life Science Laboratory National Institute of Biological Sciences

Nausea and vomiting are important defensive responses to cope with pathogens and toxins that invade the body. The nucleus of the solitary tract（NTS） is important for initiating these responses. However, the molecular heterogeneities and cellular diversities of the NTS occlude a better understanding of these defensive responses. Here, we constructed the single-nucleus transcriptomic atlas of NTS cells and found multiple populations of NTS neurons that may be involved in these defensive responses. Among these, we identified Calbindin1-positive（Calb1+） NTS neurons that are molecularly distinct from Tac1+*** Calb1+neurons are critical for nausea and retching induced by cereulide; an emetic toxin secreted by Bacillus Cereus. Strikingly, we found that cereulide can directly modulate vagal sensory neurons that innervate Calb1+NTS neurons, a novel mechanism distinct from that for nausea and retching induced by Staphylococcal enterotoxin A. Together, our transcriptomic atlas of NTS neurons and the functional analyses revealed the neural mechanism for cereulide-induced retching-like behavior. These results demonstrate the molecular and cellular complexities in the brain that underlie defensive responses to the diversities of pathogens and toxins.

关键词： vomiting nausea nucleus of the solitary tract Calbindin1 neural circuits

基于语料库的翻译文体学应用研究——以《荷塘月色》的三个英译本为例

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

大连理工大学学报（社会科学版） 2011年第4期32卷 69-73页

作者：霍跃红大连外国语学院外国语言学及应用语言学研究基地辽宁大连116044

文章选取朱自清散文名篇《荷塘月色》分别来自中国译者、西方译者和中西合作译者的三个译本为语料,利用语料库分析软件Simple Concordance Program 4.0和Textanz,从译本中的词汇变化量、句子长度、可读性,以及情感表现、比喻修辞翻译等... 详细信息

文章选取朱自清散文名篇《荷塘月色》分别来自中国译者、西方译者和中西合作译者的三个译本为语料,利用语料库分析软件Simple Concordance Program 4.0和Textanz,从译本中的词汇变化量、句子长度、可读性,以及情感表现、比喻修辞翻译等方面入手,对三个译本的文体特征进行对比分析。研究发现:三个译本在词汇层面都是低密度文本,句子都在中句范围内,与原文质朴的语言风格吻合;但在情感词汇使用、比喻修辞翻译及可读性指数上,存在着一些区别。

关键词：翻译文体学语料库应用翻译

基于CDIO教育理念的课程考核评估方法研究——以军用车辆故障诊断与排除课程为例

在线全文

学校读者我要写书评

暂无评论

军事交通学院学报 2021年第5期23卷 88-92页

作者：袁小峰霍清华卢子青刘铮陆军军事交通学院汽车士官学校安徽蚌埠233011 海军士官学校基础部安徽蚌埠233001

为有效评估学员对所学课程的综合掌握程度,提升课程教学效果,在分析士官院校课程考核存在问题的基础上,针对车辆维修技术岗位特点,构建基于CDIO的士官职业教育课程考核综合评价模型,建立以学员为主体、教员为主导的考核方式,以"五力一观... 详细信息

为有效评估学员对所学课程的综合掌握程度,提升课程教学效果,在分析士官院校课程考核存在问题的基础上,针对车辆维修技术岗位特点,构建基于CDIO的士官职业教育课程考核综合评价模型,建立以学员为主体、教员为主导的考核方式,以"五力一观"为评价准则,使用以考促学的理念来指导课程考核评估,在此过程中采用过程性考核与终结性考核相结合的方式。经过实例验证表明,该模型能够有效评价学员的综合水平,促进车辆维修技术人才发展,提高部队实战化能力和打赢能力。

关键词： CDIO教育理念课程考核评估车辆维修技术