统计推断

When we teaches the Theory of Probability and Statistics, we must pay attention to the teaching of typical mathematical thinking, such as axiomatization thinking, matrix thinking, convergence according to probability, statistical deduction and so on.

在讲授《概率统计》时，要注重公理化思想、模型思想、依概率收敛、统计推断等典型思想方法的教学。

And further more it gives a proof that this optimization problem can be seen as a convex programming in exponential neural manifold.

即将神经网络看作为统计模型，使用常用的统计推断技术，通过优化学习代价函数，对网络参数进行调整，从而达到学习的目的。

When outcome variable is measured in the dichotomous form, the descriptive indices are often indicated for frequencies, percentages or proportions, correspondingly, the statistical inference methods are mainly χ2 test or the two-variable relationship analysis.

分组变量和结局变量均为二分类变量的研究在医学研究中是比较常见的，结局变量表示为二分类的形式，研究者通常采用频数、百分比或比例等统计指标描述，相应的统计推断方法主要为χ2检验或两变量之间的关联性分析。

There are totally five chapters in the paper. Chapter 1 is the introduc-tion,and extreme value theory and methods and modelling for returns fat tails are introduced systematically in Chapter 2.This chapter also presents mean excess function and de Haan's moment estimate to decide on the reasonable exceedance threshold. Chapter 3 describes in detail non-linear time series mod-els such as GARCH models and the corresponding statistical inferences,and it also gives modelling and statistical analysis methods for returns series. Chap-ter 4 uses the models in Chapter 2 to analyzes the daily returns for Shanghai complex index and Shenzhen complex index in the Shanghai and Shenzhen stock markets. Chapter 5 applies the models in Chapter 3 to the volatility analysis of the two stock indices returns.

论文总共五章：第一章是引言部分；第二章系统介绍了极值理论方法和收益率厚尾的建模方法，给出了平均剩余函数法和de Haan矩估计相结合的方法确定合理的超越门限；第三章详细论述了非线性时间序列模型GARCH类模型及其统计推断，并且给出了收益率时间序列波动的建模和统计分析方法；第四章利用第二章的模型对国内沪深两市大盘指数上证综指和深证综指日收益率数据分布的厚尾进行了实证分析；第五章将第三章的模型应用到两个股指日收益率时间序列的波动分析。

Statistical inference under order restrictions is an important area of statistical analysis and isotonic regression theory plays a key role in this field.

约束条件下的统计推断巳成为统计分析中一个重要的研究领域，而保序回归的研究又是其中之关键。

However, according to the professional design, our mathematical statistics courses are offered only to discuss statistical inference.

但按本专业的总体设计，我们的数理统计课程只讨论统计推断。

The empirical likelihood method of Owen (1988) is a nonparametric method of inference. The method gains an advantage over the classical or the modern.

经验似然是Owen(1988)提出的一种非参数统计推断方法，这一方法与经典的或现代的统计方法比较有很多突出的优点。

The difficulty in statistical inference using Bayesian approach is calculating the posterior margin distribution, in most situations, it is very difficult to do so and even the formula can not be got.

运用Bayes 统计分析的方法进行统计推断的难点在于其后验边际分布的计算，在很多情况下很难，甚至不可能得到后验边际分布的解析表示，有时虽然可能得到解析表示但是结果相当复杂，不便于应用和推广。

Second, the t distribution is widely used in applied statistics for robust statistical inferences.

第二，多元t分布在稳健统计推断等应用统计方面使用非常广泛。

Therefore, the statistical analysis of contamination data have become an important topic.

因此，关于污染数据的统计分析已发展成为统计推断一个重要专题。

This paper discusses design and realizable methods of remote test output interface from logical design angle.

本文从逻辑设计的角度讨论遥测输出接口的设计及实现方法。

This also 星体投射plies to buildings, structures and geological features.

这也适用于建筑物和地质特征。