Optimal ensemble construction via meta-evolutionary ensembles

In this paper, we propose a meta-evolutionary approach to improve on the performance of individual classifiers. In the proposed system, individual classifiers evolve, competing to correctly classify test points, and are given extra rewards for getting difficult points right. Ensembles consisting of multiple classifiers also compete for member classifiers, and are rewarded based on their predictive performance. In this way we aim to build small-sized optimal ensembles rather than form large-sized ensembles of individually-optimized classifiers. Experimental results on 15 data sets suggest that our algorithms can generate ensembles that are more effective than single classifiers and traditional ensemble methods.     

企业不但应该做大更应该做强

从入世的角度 ,分析了我国企业在重组过程中存在的问题 ,并指出了我国企业做强做大的紧迫性     

天津通信企业IT信息存储支撑平台

天津通信企业IT信息存储支撑平台承载着天津通信关键业务支撑系统和主要的企业信息管理系统．为IT运营管理提供了高效、快速、稳定、安全的数据源，形成统一的存储平台。该平台采用了多系统存储资源集中共享的方式，通过海量数据的异构存储平台数据迁移实现了多系统的整合．达到多系统的数据资源共享。存储的整合为应用提供了可靠的资源保证，数据的整合则消除了信息孤岛，实现了数据共享。     

A new iterative order reduction (IOR) method for eigensolutions of large structures

Order reduction is a computationally efficient method to estimate some lowest eigenvalues and the corresponding eigenvectors of large structural systems by reducing the order of the original model to a smaller one. But its accuracy is limited to a small range of frequencies that depends on the selection of the retained degrees of freedom. This paper proposes a new iterative order reduction (IOR) technique to obtain accurately the eigensolutions of large structural systems. The technique retains all the inertia terms associated with the removed degrees of freedom. This hence leads to the reduced mass matrix being in an iterated form and the reduced stiffness matrix constant. From these mass and stiffness matrices, the eigensolutions of the reduced system can be obtained iteratively. On convergence the reduced system reproduces the eigensolutions of the original structure. A proof of the convergence property is also presented. Applications of the method to a practical GARTEUR structure as well as a plate have demonstrated that the proposed method is comparable to the commonly used Subspace Iteration method in terms of numerical accuracy. Moreover, it has been found that the proposed method is computationally more efficient than the Subspace Iteration method. Copyright © 2003 John Wiley & Sons, Ltd.     

浅议铵梯炸药质量检验中的三个问题

文章探讨了GB 12437-2000标准规定的产品检验项目中,中包浸水试验的科学性问题和对批产品进行抽样时的样本代表性问题以及使用多段爆速仪测试爆速时应注意的样本量问题.     

Influence of nano-scaled dispersed second phase on substructure of deformed dispersion strengthened copper alloy

The deformation behavior of dispersion strengthened copper alloy Cu-Al2O3 was studied by TEM. The results show that nano-scaled dispersed second phase not only increases dislocation density in matrix, but also has an important influence on the dislocation substructure. The presence of fine dispersed Al2 O3 particles results in a uniform and random dislocation distribution in matrix copper and causes the difficulty in formation of dislocation cell structure and the decrease in the amount of cell structure during deformation. Deformation gives rise to much more dislocations and dislocation cells form more difficultly and the decrease in the cell size with the increase of dispersion degree.     

纳米颗粒的测量与表征

介绍了用于纳米颗粒测量的电镜观察法、X射线衍射线宽法、激光粒度分析法、比表面积法、颗粒沉降法、扫描探针显微术以及小角X射线散射等,并对其测量原理、测量过程、适用范围及测量方法的优缺点进行了讨论。     

Terminal velocity of non-spherical particles falling through a Carreau model liquid

A procedure has been proposed for the estimation of the terminal falling velocity of non-spherical particles moving in a Carreau model fluid in the transition flow region. The procedure is based on a modification of the relationship formerly developed for the fall of spherical particles including the particle dynamic shape factor. The suitability of the proposed procedure has been confirmed by good agreement between experimental and calculated terminal falling velocity data. In the experiments, the terminal falling velocity of short cylinders and rectangular prisms in polymer solutions of different measure of shear thinning and elasticity has been measured.     

大规模训练数据的支持向量机学习新方法

支持向量机不能直接对大规模的训练数据进行学习。提出一种新的减小样本集规模的方法；在映射后的高维空间中寻找两种类别的交界部分，交界部分上的样本作为学习样本。并且指出，不需涉及具体映射的形式只用核函数即可找到交界部分的样本。实验表明，新方法优于直接在低维样本空间中寻找交界部分样本的方法。     

Alpha induced reactions on Sn: An experimental study of excitation functions and possible production pathways

In the frame of a systematic study of light ion induced threshold reactions on natural tin (Sn) excitation functions for alpha particles induced reactions are presented in a 12–38 MeV energy domain. Using a stacked foil activation method the following radioisotopes were identified: ¹¹⁶Te, ¹¹⁷Te, ¹¹⁸Te, ¹¹⁹Te, ¹²¹Te, ¹²³Te, ¹¹⁷Sb, ¹¹⁸Sb, ¹²⁰Sb, ¹²²Sb, ¹²⁴Sb, ¹²⁶Sb, ¹¹⁷Sn, ¹¹¹In. The experimental cross sections for these isotopes are presented for the first time in this energy range and a direct comparison with values calculated with the ALICE-IPPE code is discussed. Possible use of these data for production of some isotopes relevant in nuclear medicine is suggested.