后验

A method to make Credal network approximation inference is proposed by computing the joint probability distribution and posterior probability of variables through the above operators.Furthermore its arithmetic is developed by SQL statement.Its validity and efficiency can be showed in the application case.

提出利用上述操作计算Credal网络中各结点联合概率分布和后验概率的方法，从而实现了Credal网络的近似推理，并给出了该近似推理的SQL语言实现算法，算例结果表明该方法是有效的。

The inference in Credal network is the computation of tight lower and upper bounds for conditional probabilities.A new algorithm for inference in Credal networks is presented based on the framework of bucket elimination.The burden of computation is alleviated by enumerating some of values and the accurate result is obtained.

Credal网络的推理是计算一定证据下的后验概率最大值和最小值，给出了一种Credal网络推理的新方法，该方法是在桶消元框架下通过枚举计算部分因子函数值，使计算量大大减小，并且可以得到精确的结果。

In view of the peaked and fat-tailed characteristics of financial return data distribution and its effect of clustering fluctuation and especially the "leverage effect" of fluctuation on VaR estimates and some efficiencies when estimating VaR with various assumptions of return data distribution,a semi-parameter approach based on EGARCH-VaR model is developed.

在综合考虑了金融收益数据分布的尖峰厚尾特征及其波动集群性，尤其是其波动的&杠杆效应&对VaR估计的影响以及各种假定收益率分布在计算风险价值时存在不足的基础上，提出了基于EGARCH-VaR的半参数方法，并且与正态分布和t分布假设下的GARCH模型的VaR计量方法进行比较，通过实证分析，并利用后验测试，表明基于EGARCH-VaR的半参数方法对风险价值的测度优于正态分布和t分布假设下GARCH模型的VaR计量方法。

Results Posterior medians and 95 percent equally tailed posterior credible intervals were obtained.

结果 可计算出后验参数估值及其95% Bayesian可信区间。

Monte Carlo is a method that approximately solves mathematic or physical problems by statistical sampling theory. When comes to Bayesian classification, it firstly gets the conditional probability distribution of the unlabelled classes based on the known prior probability. Then, it uses some kind of sampler to get the stochastic data that satisfy the distribution as noted just before one by one. At last, it can obtain the posterior probability distibution of each unlabelled classes by analysing these stochastic data.

蒙特卡罗是一种采用统计抽样理论近似求解数学或物理问题的方法，它在用于解决贝叶斯分类时，首先根据已知的先验概率获得各个类标号未知类的条件概率分布，然后利用某种抽样器，分别得到满足这些条件分布的随机数据，最后统计这些随机数据，就可以得到各个类标号未知类的后验概率分布。

When this procedure is repeated, the impact of samples is more and more important, so the assessment of priors is vital to the efficiency and quality of learning.

Bayesian方法既可避免只使用先验信息可能带来的主观偏见，和缺乏样本信息时的大量盲目搜索，也可避免只使用后验信息带来的噪音的影响。

Key words: speaker verification; GMM; UBM; MAP

说话人确认；高斯混合模型；背景模型；最大后验概率算法

Firstly, the features were divided into L groups and the SVM hyperplanes were constructed for each feature of training set. Secondly, the testing set was tested by SVMKNN method, and the decision profile matrixes were obtained. Finally, these decision profile matrixes were implemented by multi-feature fusion method.

在该算法中，设样本集特征可分为L组，先用SVM算法根据训练集中每组特征数据构造分类超平面，共构造L个；其次用SVMKNN方法对测试集进行测试，得到由L组后验概率构成的决策轮廓矩阵；最后将其进行多特征融合，输出最终的分类结果。

Under some suitable conditions, a priori estimate and a posteriori one are given.

在适当的条件下，得出该算法的先验与后验误差估计。

In this thesis, we study the problem about subconvergence and a self-adaptive posteriori error estimates of bilinear finite element.

中文摘要：本文主要是研究双线性有限元的慢收敛和自适应后验误差估计的有关问题。

With Death guitarist Schuldiner adopting vocal duties, the band made a major impact on the scene.

随着死亡的吉他手Schuldiner接受主唱的职务，乐队在现实中树立了重要的影响。

But he could still end up breakfasting on Swiss-government issue muesli because all six are accused of nicking around 45 million pounds they should have paid to FIFA.

不过他最后仍有可能沦为瑞士政府&议事餐桌&上的一道早餐，因为这所有六个人都被指控把本应支付给国际足联的大约4500万英镑骗了个精光。