可并行中文同主题词聚类新算法

提出了一种高效的自动按照主题对中文词进行聚类的算法.该算法利用顿号(、)切分抽取语料库句子中的并列中文词,并以抽取出的中文词为节点构建一个共引用图; 然后对每个中文词节点产生若干个locality sensitive Hashing (LSH)签名组合; 最后将至少有1个相同LSH签名组合的任意2个中文词标记为同一个主题类.实验表明,该算法运算速度快,且易并行实现,在海量语料库的支持下,执行效率高,聚类效果较好.     

基于深度相机的手腕识别与掌心估测算法(HHME2013-ID51)

当手和手臂都进入深度相机所设定的有效深度范围时,它们将被作为一个整体来提取,若处理时也把它们作为一个整体,这可能会影响手势交互的一些重要算法,如掌心估测,手朝向估测,手的跟踪等。本文分析了手腕的运动特征和手的轮廓特点,并利用内切矩形的几何特征,提出了手腕识别算法来分割手与手臂;为了提高掌心估测算法的性能,本文分析了锐角三角形和最大内切圆的几何特征,并结合手势交互的特点,提出了新的掌心估测算法。经实验证明,手腕识别算法能较好地分割出手与手臂,新的掌心估测算法较之原算法在性能上有显著提高。     

一种改进的智能卡数据传输安全策略

针对传统智能卡进行数据传输时仅支持单一模式的缺点,提出一种改进的基于模式控制字的数据传输安全策略。此模式控制字支持多种数据传输模式,数据可以采用明文、密文、明文MAC、密文MAC中的任一种方式进行传输,并通过基本文件(包含透明文件、线性文件)的读、写模式控制字,来规定文件进行读、写操作需要满足的传输安全条件。同时,还提出使用禁止添加、或禁止更新这两种写属性来限制基本文件的写入方式。用户可以根据自己的需求,将同一文件的读命令,和写命令采用的传输方式设置成不一样,从而选择不同的数据传输方式,进而提高数据传输的灵活性和安全性。     

也谈士官教育——Word一堂课教学设计

熟练掌握office办公软件,是士官学员毕业后参与部队训练、管理、教育必须掌握的技能。为了培养士官学员的这一技能,针对士官职业技术教育中实施信息技术基础教学既要引导学员对办公自动化软件基本功能的熟悉,又要有效指导学员具体实践操作,还要在所学知识的基础上能有拓展提高的要求,在Word一堂课教学实施过程中,可以采取任务驱动,能力训练项目来展开教学,以达到理想的成效。     

Root growth and phosphorus and potassium uptake by two corn genotypes in the field

In a field study P and K uptake by two corn (Zea mays L.) genotypes which differed in root growth was investigated. The effect of differences in root growth on P and K uptake was assessed using a mechanistic-mathematical model which describes nutrient uptake by growing plant roots in soil. Nitrogen was applied at 0 and 227 kg ha^–1 to Pioneer 3732 and B73xMo17 corn grown on Raub silt loam (fine-silty, mixed, mesic Aquic Argiudoll) and at 227 kg N ha^–1 to these two genotypes on Chalmers silt loam (finesilty, mixed, mesic, Typic Haplaquoll). Root growth and P and K uptake by the two corn genotypes was measured 31, 47, 75 and 91 d after planting on the Raub and 31, 47, 61 and 75 d after planting on the Chalmers soil.Root growth and P and K uptake by B73xMo17 was greater than that of Pioneer 3732 on N-fertilized Raub soil. On Chalmers soil the difference in root growth between the two genotypes resulted in an increase in K but not P uptake. The higher soil P level of the Chalmers appears to have offset possible differences in P uptake due to root size. There were no differences between the two genotypes in either the percentage of roots with root hairs, or the density or length of root hairs. Phosphorus and K uptake calculated with the simulation models for both corn genotypes on both soils over each of three growth periods agreed with observed P (Y = 0.68X + 1.71; r = 0.944^**) and K (Y = 0.88X + 15.52; r = 0.928^**) uptake. Differences in P and K uptake between B73xMo17 and Pioneer 3732 resulted primarily from the difference in root growth in the topsoil. A high correlation was found between root surface area and P (r = 0.893^**) and K (r = 0.928^**) uptake by both corn genotypes on both the soils.Journal paper No. 10,316 Purdue Univ. Agric. Exp. Stn., W. Lafayette, In 47906. Contribution from the Dep. of Agron. This paper was supported in part by a grant from the Tennessee Valley Authority.     

Inhibition and recovery of cucumber roots given multiple treatments of ferulic acid in nutrient culture

Ferulic acid, a frequently cited allelopathic agent, inhibited photosynthesis, leaf expansion, and root elongation of cucumber seedlings grown in aerated nutrient cultures in a growth chamber. Other effects were a reduction in the proportion of radioactivity fixed by photosynthesis translocated to roots, a stimulation in secondary root initiation, and an increase in root-shoot ratios. Inhibition of leaf expansion and root elongation induced by multiple ferulic acid treatments was rapidly lost once ferulic acid was removed from the root environment. The changes in general root morphology, i.e., average root length and root number, associated with ferulic acid treatments, were partially reversed or not affected when ferulic acid was removed from the root environment.Paper No. 11411 of the Journal Series of the North Carolina Agricultural Research Service, Raleigh, North Carolina 27695-7601. The use of trade names in this publication does not imply endorsement by the North Carolina Agricultural Research Service of the products named, nor criticism of similar ones not mentioned.     

一种基于决策树和词义相似度的N1+N2结构语法关系判定方法

设计了一种基于决策树算法的N1+N2结构语法关系判定方法.首先建立了该结构的语料库,对每条语料都标注构建特征集所需的词性、《同义词词林》语义编码、语法关系和词义相似度等信息;然后为证明相似度在判断语法关系中的合理性,根据语言学原理研究了N1+N2结构两个名词间语义相似度与语法关系之间的内在规律:①从语法关系的角度比较两个名词间的语义相似度结果为:并列关系>复指关系>定中关系>主谓关系;②从语言功能焦点的角度比较两个名词间的语义相似度结果为:并焦型短语>后焦型短语.最后以此为基础构建了特征集,运用决策树C4.5算法建立了自动判定N1+N2结构语法关系的方法.运用该算法在自建语料库的测试集中进行测试,正确率为89.39％.     

基于信道状态特征的手势动作识别技术研究

手势识别作为人和机器之间重要的交互手段,在日常生活中具有广泛的应用场景。基于无线信号特别是WiFi的手势识别由于其无接触、成本低等优点成为当前热门的方式。为解决传统基于无线信号手势识别算法没有充分利用信号相位特征的缺点,本文提出利用WiFi信道状态信息幅值和相位结合的方式进行手势识别。通过子载波降维和动作曲线提取对接收的WiFi信号进行处理,并将信号的幅值和相位结合,利用机器学习算法对数据进行训练和分类,实现了单手向前、单手向后、单手向左、单手向右、单手向上和单手向下六种手势的识别。实验结果显示,本文算法在近距离和远距离下的精确度分别为96%和92%。     

基于稳健设计的SGNS算法的超参数调优方法

通常给定超参数的若干取值选取性能最大的为最优组合(称为直接选优法),但是此方法的稳健性差。为此,提出了一种基于稳健设计的超参数调优方法(称为稳健调优法)。具体地,以SGNS算法中的超参数调优为例,在词语推断任务上实验并得出:经方差分析得到SGNS算法中的七个超参数中的五个对算法预测性能有显著影响,确定为主控因子,其余两个确定为噪声因子,且主控因子中有三个对性能估计的方差有显著影响,因此,调优中仅从期望最大来直接选优是不合理的;稳健调优法与直接选优法两者在预测性能上没有显著差异,但稳健调优法对噪声因子具有较好的稳健性。稳健调优法对一般的深度神经网络的调参有实际的借鉴意义。     

基于轻量级OpenPose改进的幻影机手势交互系统

目前人机交互方式多以键盘鼠标为主,而基于深度学习手势识别的交互方式算法准确率不高,且实时性和系统稳定性均有待提升。提出一种新颖的针对轻量级OpenPose进行改进的幻影机手势交互系统。采用轻量级OpenPose将人手简化建模为21个关键点,以MobileNetV1作为基础模型,应用部分亲和域（Part Affinity Fields,PAF）方法实现人手关键点的检测并画出简化骨骼图。为进一步提升人机交互系统的实时性,采用幻影模块（Ghost Module）对卷积层进行降维,用更少的硬件资源取得同样的识别效果。最后搭建验证环境,根据画出的人手骨骼图进行模式匹配,根据匹配识别结果生成交互控制指令,经由蓝牙通讯将指令传送至Arduino UNO平台控制小车实现交互响应。经过初步训练后,该系统在COCO2017验证集上能实现58.7%的准确率,保持了原始OpenPose网络和轻量OpenPose网络的人手关键点识别效果,在家用PC机上可实现每秒32~36帧的识别速率和较高的手势识别率。