Extensions of LDA by PCA mixture model and class-wise features

Linear discriminant analysis (LDA) is a data discrimination technique that seeks transformation to maximize the ratio of the between-class scatter and the within-class scatter. While it has been successfully applied to several applications, it has two limitations, both concerning the underfitting problem. First, it fails to discriminate data with complex distributions since all data in each class are assumed to be distributed in the Gaussian manner. Second, it can lose class-wise information, since it produces only one transformation over the entire range of classes. We propose three extensions of LDA to overcome the above problems. The first extension overcomes the first problem by modelling the within-class scatter using a PCA mixture model that can represent more complex distribution. The second extension overcomes the second problem by taking different transformation for each class in order to provide class-wise features. The third extension combines these two modifications by representing each class in terms of the PCA mixture model and taking different transformation for each mixture component. It is shown that all our proposed extensions of LDA outperform LDA concerning classification errors for synthetic data classification, hand-written digit recognition, and alphabet recognition.     

论科技论文英语摘要的表现形式

论述科技论文中英语摘要的文体风格、时态、语态和基本要求，并结合例句介绍了常用的表现形式，使之简洁、准确和流畅。     

Distinctiveness and Correlation in Conceptual Structure: Behavioral and Computational Studies.

Patients with category-specific deficits have motivated a range of hypotheses about the structure of the conceptual system. One class of models claims that apparent category dissociations emerge from the internal structure of concepts rather than fractionation of the system into separate substores. This account claims that distinctive properties of concepts in the living domain are vulnerable because of their weak correlation with other features. Given the assumption that mutual activation among correlated properties produces faster activation in the normal system, the authors predicted a disadvantage for the distinctive features of living things for unimpaired adults. Results of a speeded feature verification study supported this prediction, as did a computational simulation in which networks mapped from orthography to semantics. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Probabilistic analysis of uncertainties in seismic hazard assessment

Sources and current methods of analysis of uncertainty from randomness, fuzziness and ignorance or incomplete knowledge in seismic hazard assessment problem are briefly discussed at beginning; understandings of the authors are then presented in the following order. All three types of uncertainty come from incomplete knowledge. Probabilistic method can be applied to all of them, objective probability for random factors and subjective probability for the other two types of uncertain factors. Discrete subjective probability mass functions for incomplete and fuzzy factors can be obtained from logic-tree and membership functions respectively. Fractile curves may be used to show the scattering of any uncertainty factor, but a unified probabilistic treatment may be applied to any combination of all three types of uncertainty.     

北京火车站抗震鉴定与加固技术

北京火车站是首都重要的标志性建筑之一 ,195 9年建成 ,当时按 7度设防考虑 ,存在着许多不符合抗震要求的问题 ,一些设施也明显不能满足现代使用功能要求。北京火车站抗震加固与改造设计项目组 ,运用多种先进的加固手段 ,在保持原建筑外貌不变的前提下 ,实现了抗震能力和改善使用功能的协调 ,取得了大型公共建筑加固改造的新经验     

挡土墙填土曲线破裂面的地震主动土压力分析

在地震作用影响下，墙后填土处于极限平衡状态时，假定破裂面为一条通过墙趾的旋轮线。从墙后土楔中取出一微薄水平层，研究其在水平惯性力下的平衡，得到的地震土压力分布和土压力合力作用点位置均与实验结果相吻合。     

地震复合波地质属性的研究方法

在地震资料解释中,地震复合波的地质属性问题一直没有得到很好地解决。本文针对松辽盆地砂泥岩薄互层的地质特点,提出了用波形合成追踪法研究地震复合波地质属性的方法。该方法在地震复合波正演合成过程中有五个可以识别的特征界面,其中 C 界面是地震复合波形成的主导界面;C 至 D 界面间的地层为形成地震复合波的主要地层。因此,找准 C 界面深度是建立地震复合波与钻孔地层间有机联系的关键,对地震地质层位准确标定和利用地震剖面复合波波形信息进行岩性预测极为重要。文中的应用实例表明,用该方法在松辽盆地较好地建立了地震信息与地质信息间的对应关系,解决了一些疑难地质问题,取得了较好的勘探效果。     

叠后记录Radon变换压制干扰方法

在叠加剖面中,下伏反射界面的反射信号同相轴呈连续的或带有限个间断点的连续层状图像,各种干扰则呈无规则的图像。对此剖面的振幅数据进行 Radon 变换,即分别向各方向进行投影,层状图像将产生一定形状的光滑曲线,而干扰数据将相互抵消,其剩余数据呈高斯分布。在Radon 域中通过平滑等方法消除这些呈高斯分布的噪声数据,再经 Radon 逆变换,便可获得信噪比较高的叠加剖面。其结果不改变原叠加剖面的分辨率和信号波形特征。理论与实际剖面的处理结果表明,这是一种良好的压制干扰方法。     

防晃水箱控制高层建筑地震反应的理论研究

防晃水箱是在一般水箱内安置环形防晃板，在高层建筑顶部设置这种水箱可以对建筑物的地震反应作被动控制。本文对这一系统进行了数值模拟，考虑了水箱和结构间的相互耦合作用，研究了水箱对建筑物地震反应的影响，数值分析为合理设计水箱提供了理论依据。