基于密度优化的KNN算法的研究

通过对样本网页文本的影响因子特征提取,构建向量空间模型,同时利用OPTICS算法密度无关性,改进了KNN算法.实验表明,该算法增强了结果的稳定性,并产生质量较高的聚类结果.     

SG-VMD-SVD的信号去噪方法研究

油气管道信号泄漏检测易受噪声影响,因此去噪成了关键问题.为了提高对油气管道信号的去噪效果,提出了一种基于Savitzky-Golay平滑滤波、变分模态分解(VMD: Variational Mode Decomposition)和频域奇异值分解(SVD:Singular Value Decomposition)去噪相结...     

分离点已知的二维水翼超空泡形状研究

采用奇异积分方程方法,研究了二维薄翼定常超空泡问题．应用一种改进的离散化方法对积分方程进行了求解．当超空泡在水翼上下表面的分离点已知时,给出了超空泡形状的．给出了几种非对称翼型超空泡形状的数值结果．     

一类二阶奇异边值问题解的存在性及其解的性质

研究了二阶奇异线性边值问题{Ф″（t）＋n-1/tФ′（t）＋λФ（t）=0,0＜1＜1,;Ф′（0）=0,Ф（1）=1,利用Frobenius方法得到了该问题的解的存在性,并给出了其解的幂级数表达式和这个解的一些性质（n≥2,λ≠0）．     

Does democracy require value-neutral science? Analyzing the legitimacy of scientific information in the political sphere

Philosophers now commonly reject the value free ideal for science by arguing that non-epistemic values, including personal or social values, are permissible within the core of scientific research. However, little attention has been paid to the normative political consequences of this position. This paper explores these consequences and shows how political theory is fruitful for proceeding in a world without value-neutral science. I draw attention to an oft-overlooked argument employed by proponents of the value free ideal I dub the “political legitimacy argument.” This argument claims that the value-free ideal follows directly from the foundational principles of liberal democracy. If so, then the use of value-laden scientific information within democratic decision making would be illegitimate on purely political grounds. Despite highlighting this unaddressed and important argument, I show how it can be rejected. By appealing to deliberative democratic theory, I demonstrate scientific information can be value-laden and politically legitimate. The deliberative democratic account I develop is well suited for capturing the intuitions of many opponents of the value free ideal and points to a new set of questions for those interested in values in science.     

How to make value-driven climate science for policy more ethical

In previous works, I examine inferential methods employed in Probabilistic Weather Event Attribution studies (PEAs), and explored various ways they can be used to aid in climate policy decisions and decision-making about climate justice issues. This paper evaluates limitations of PEAs and considers how PEA researchers’ attributions of “liability” to specific countries for specific extreme weather events could be made more ethical. In sum, I show that it is routinely presupposed that PEA methods are not prone to inductive risks and presuppose that PEA researchers thus have no epistemic consequences or responsibilities for their attributions of liability. I argue that although PEAs are nevertheless crucially useful for practical decision-making, the attributions of liability made by PEA researchers are in fact prone to indicative risks and are influenced by non-epistemic values that PEA researchers should make transparent to make such studies more ethical. Finally, I outline possible normative approaches for making sciences, including PEAs, more ethical; and discuss implications of my arguments for the ongoing debate about how PEAs should guide climate policy and relevant legal decisions.     

多效唑和摘心对百日草观赏效果的影响

探讨了多效唑、摘心处理对百日草观赏效果的影响,结果表明,摘心和多效唑对百日草均有延迟开花、节间缩短、植株矮化、株型紧凑、花期延长的作用,其中以二次摘心开花最迟,浓度为1 000 m g/L的15%多效唑处理观赏效果最佳。     

P2P系统分布式LSI的构建和更新

从P2P系统自组织和动态性特点出发,提出分布式环境下隐语义索引(LSI)构建和更新的P2P网络模型,设计适合P2P系统文档矩阵的降维表示(RDR)合并算法,结合信号和噪声子空间模型从理论上分析RDR合并算法的有效性及算法需要满足的前提条件;使用M atlab6.5针对标准文集测试RDR合并算法对查询精度的影响.理论分析和数字实验证明,该算法能够解决P2P系统中分布式LSI的构建和更新问题,能在可容忍的查询精度影响范围内,以较低的网络开销和计算量分布式地构建、更新隐语义索引.     

半线性抛物方程组的猝灭时间与猝灭速率估计

考虑含有奇异项的半线性抛物型方程组的初边值问题,证明了当区域Ω适当大,使在Ω上L ap lace算子在齐次D irichket边界条件下特征值问题的第一特征值小于某一常数时解会在有限时刻发生猝灭,并对猝灭时刻的上、下限进行估计,其次对猝灭速率进行讨论。     

参数曲面奇点的性质

从参数曲面奇点的定义出发,给出了参数曲面上奇点成立的条件,并得到了参数曲面奇点处的第一基本形式、第二基本形式及高斯曲率。在此基础上,找了Bézier曲线曲面奇点产生的原因。