查询词典 iterated kernel

In this paper, the performance of SVM based on Fourier kernel is studied which aims at the regression in signal processing questions, and the influence of parameter q on performance of SVM is analyzed. A conclusion is drew that the integral of Fourier kernel in one period is a constant and the concept of equivalent kernel function width is proposed.

本文针对信号处理中的回归问题，研究了基于傅立叶核函数支持向量机的性能，并分析了参数q对其性能的影响；根据推导得出傅立叶核函数在单个周期内积分为常数的结论，并提出了等效核宽度的概念；最后的仿真结果证实了，在信号处理领域，采用傅立叶核函数的支持向量机比采用径向基核函数的有着更好的性能。

Inspired by the intuitive geometric interpretation of SVM based on convex hulls, KNNCH maps the data in the original space to the kernel space with the kernel trick and constructs a nearest neighbor classifier in the kernel space, which takes the convex hulls of training sets as the extended class sets.

该算法的设计受到支持向量机几何解释启发，利用核函数方法将数据映射到高维核空间，并在核空间构造以训练集凸包为扩展类集的最近邻分类器。

Firstly, Wavelet-Galerkin algorithm for solving the first kind of singular integral equation with the Hilbert kernel is proposed, we use the characteristic of periodic wavelet on L~2([0,1]) and Hilbert kernel to solve and make stiff matrix lower dimensions and become sparser through thresholding,thus the cost of computation is reduced. Because of the singularity of Hilbert kernel we use Tikhonov regularization method to solve the system of stiff equation. At last the convergence and numerical result of approximate solution are given. Secondly, an approach of regularization based on Fourier is presented for sideways heat equation; we give the theory proof and error estimate.

首先，提出了含Hilbert核的第一类奇异积分方程的小波伽辽金(Wavelet-Galerkin)数值算法，该算法中利用了L~2([0，1])上的周期小波和Hilbert核的特点进行处理，使得刚性矩阵维数降低并且通过阈值使得它更加稀疏，减少了计算量；由于Hilbert核的奇异性，通过Tikhonov正则化方法求解所得到的刚性方程组，给出了收敛性和数值结果；其次，对标准的一维逆热传导方程给出了一种基于Fourier正则化方法，给出了理论证明及其误差估计，解决了文献中算法与理论误差估计的不相匹配的现象，该正则化方法不仅保留了测量数据的部分高频成份，且与文献中的算法具有同样的计算量和误差估计。

Numerical results show that constant coagulation kernel has more influences on small particles than linear and quadratic coagulation kernel,whereas has less influences on big particles than linear and quadratic coagulation kernel;in the same wa...

发现常凝并核要比线性和二次方凝并核对小颗粒的影响大一些、对大颗粒的影响小一些，线性和二次方凝并核对颗粒尺度分布的时间演变影响则取决于具体情况；常冷凝核要比线性冷凝核对小颗粒的影响大一些、对大颗粒的影响小一些；连续区布朗凝并核类似于常凝并核；扩散冷凝核对颗粒尺度分布的影响界于常冷凝核和线性冷凝核之间。

Topics covered include: density estimation (incorporating such topics as kernel density estimates, nearest neighbor estimates, variable kernel methods, orthogonal series estimates, maximum penalized likelihood estimates, bandwidth choice), nonparametric regression (incorporating kernel methods, spline methods, penalized likelihood, bandwidth choice, generalized additive models), and bootstrapping.

非参数建模专题：非参数建模专题：基于计算机的非参数统计方法论。包括：密度估计(如：核密度估计，最近邻估计，可变核方法，正交序列估计，最大补偿似然估计法，带宽选择)非参数回归(如：，核方法，样条方法，补偿似然估计，带宽选择，广义可加模型，)和 bootstrapping 方法。

Firstly, a two-dimension tensor product space is constructed, in which reproducing kernel exists. Secondly a multiresolution analysis is constructed in the reproducing kernel space. Then an orthonormal basis in is obtained. Thus,the reproducing kernel space can be expressed by wavelets spaces. Wavelets approximation formula and sampling formula can also be obtained in the space. Besides, the wavelets approximation formula is simple and easy for calculation. That completes the theory of multiresolution analysis in finite interval fully and provides theoretical base and algorithm for application.

首先，构造二维张量积空间，并证明该空间具有再生核；其次在再生核空间中，建立多尺度分析，获得该空间的一个标准正交基，使得再生核空间可以由小波空间来刻画，进而得到小波级数和相应的采样公式，而且给出的小波级数形式简单易于数值分析，进一步完善了有限区间上的多尺度分析方法，也为实际应用提供了良好的理论基础和算法。

The method improves performance of KPCA by reasonably combining a local conditionally positive definite kernel and a global kernel. The experiment results indicate the new kernel function certainly improves the performance of KPCA.

利用组合核函数提高核主分量分析的性能，该方法将一个全局核函数与一个新的局部核函数条件正定核函数进行合理组合，从而提高KPCA所提取特征的质量。

Through analyses, we can make the conclusion that, the relation of seed fresh weight dry weight and days after pollination, fat percent content, protein nut content, starch nut content can be expressed with Logistic equation; the change of kernel volume, moisture content of kernel and days after pollination accorded with three ascend curve equation. Kernel protein content, starch content and the changes of days after pollination tallied with reciprocal equation and they had high relation.

经过统计与分析，可以看出籽粒鲜重积累量与授粉后时间、籽粒干重与授粉后的时间、脂肪净含量、脂肪百分含量、蛋白质净含量、淀粉净含量与授粉后的时间均可以用Logistic方程描述，籽粒体积变化与授粉后的时间关系表现为三次上升型曲线方程；籽粒的水分含量与授粉后的时间关系表现为三次下降型曲线方程；籽粒中蛋白质百分含量与授粉后时间的动态变化规律、籽粒中淀粉百分含量与授粉后时间的动态变化规律符合倒数方程，均具有较高的拟合度。

By using testing functions to make comparative research on the precision of approximate models which were built by RBF under the condition of different sample amounts and different kernel function types, a conclusion was drawn: the model quality built by Reciprocal Multiquadric RBF as kernel function was highest when the sample amout was big, and the model quality built by Gaussian RBF as kernel function was highest when the sample amount was small.

运用测试函数对在不同样本数量与不同核函数形式下所建立的近似模型的精度进行对比研究，得出逆多二次函数与高斯函数作为核函数分别在大样本和小样本情况下精度最高的结论，并验证了径向基函数近似模型方法能够兼顾效率与精度。

Then, we focus on the construction of a new type of kernels. It has been pointed out that the kernel function of SVM is equivalent to the reproducing kernel of a certain Reproducing Kernel Hilbert Space.

首先对线性核函数、Gaussian核函数等已经用于广泛实践的核函数的性能给出了简要评述，然后重点构造了一类新的核函数。

Nowadays, most of research are to build a transmutative Petri Nets through adding controlling place sets, controlling arc sets and controlling policy to the basic Petri Nets, while the Controlled Petri Nets could be used to argue many controlling theory problems conveniently and to induce many logically and physically supervisory and solve the Event Feedback Controlling Problems and State Feedback Controlling Problem in DEDS supervisory theory.

目前大多数的研究表现为在变形后的受控Petri网基础上，利用各种方法求得各种逻辑型、结构型控制器，解决DEDS监控理论中的事件反馈控制问题与状态反馈控制问题。

On one hand, there are discussions with the works council and union about extension of short time working up to the end of September.

一方面，有讨论，工程理事会和联盟关于延长工作时间短至9月底。