值

Thirdly,quasi-normal inequalities of a-variation maximal operator and a-conditional variation maximal operator of scalar predictable tree martingales areidentified by the use of martingale transforms and by the construction of convex or concave function method;on this basis and with the help of previsiblity or regu-larity,Burkholder-Davis-Gundy\'s inequality of a-variation maximal operator and a-conditional variation maximal operator of scalar predictable tree martingales are iden-tified by the application of Hardy-Lorentz interpolation theory.At the same time,bythe use of G.

再次，应用鞅变换和构造凸或凹函数方法证明了标量值可料树鞅的α-方极大算子和α-条件方极大算子的拟范数不等式；然后，在这些拟范数不等式的基础上，应用Hardy-Lorentz空间插值方法证明了当树鞅是可控或正规树鞅时关于标量值可料树鞅α-方极大算子和α-条件方极大算子的Burkholder-Davis-Gundy's不等式成立。

The value of a conditional expression is defined as follows: If the effective boolean value of the test expression is true, the value of the then-expression is returned.

一个条件表达式的值定义如下：如果测试表达式的有效布尔值为true，则返回then表达式的值。

Then, based on conservation of energy law, the physical cause of it was analyzed.

然后根据能量守恒原理，分析了混凝土浇筑结合面上节点初始温度取值不同时，造成温度场计算值与理论分析值偏差的物理原因。

Thiele type continued fraction interpolation is used to interpolate pixels around damaged part of the regular texture aiming at repairing the texture based on the regular inpainting method. Bisection is used when whole information and details are needed in sampling. Thiele type continued fraction interpolation is not an ideal method in the case of irregular damaged region Newton-Thiele type contined fraction is put forword to solve this problem. Point window is also used in the process of sampling.

在通常的图像修复技术上，对于一些规则的破损纹理图案采用一元Thiele型连分式模型来进行修复，在采样选取上，考虑到图像的整体信息和细节特征采用了二分的方法来进行处理，对于具有不规则破损区域的纹理图案，采用一元的方法不能达到令人满意的效果，采用二元Newton-Thiele型混合有理插值连分式的有理插值的方法来对具有不规则破损部分周围的像素点进行插值并采用采样点窗的方法进行采样可以达到修补破损部分的目的。

However,the threshold of traditional threshold method is simplex,and in the condition of poor continuity of the signal,the use of this method will produce pseudo-Gibbs phenomenon.

但传统的阈值法阈值单一，且在信号连续性较差的情况下，运用传统阈值法还会产生伪吉布斯现象。

The relationship between the topology and convegence of networks is approached and the reason causing overtraining and overfitting is analysed. Then the redundant weights and nodes in the network are defined by using improved activity function and a new algorithm called the BPNSO for determining the network structure for any application is proposed, which use dual optimization objective function combining the mean squre error of the network output with network complexity penalty function.

文章对网络拓扑结构与网络收敛性和稳定性的关系进行了探讨，分析了网络出现Overtraining或Overfitting的原因和解决的目标，提出了神经网络结构优化的策略，改进了活化函数，定义了冗余权值和冗余节点的概念，为辨识和删除冗余权值和冗余节点，提出采用优化目标函数中增加网络复杂度罚函数的双重优化目标函数，并对罚因子提出了基于学习误差和权值的自适应动态调整公式，论文定义了这种BP网络结构优化方法为BPNSO，并给出了完整的计算流程。

From our experimental results, we find the expanding input features with back propagation learning rule can get fast convergence in training and decrease the mean absolute error between the desired output and the actual output.

实验显示，采用此种扩展型特徵值的倒传递学习法网路可以加速网路的学习效率以及降低实际输出值与期望输出值之间的绝对值误差。

By analysis of such models, a method, which is used to find out the region of convergence and to study the local stability for strong nonlinear system, are derived. Thus, we can consider the strong nonlinear neural networks using such method.

利用新的比较方法，考虑了具变时滞和常恢复率的同质人群HIV梯度传染病模型，得到了一个阈值，当阈值小于1时，疾病消除平衡点全局指数渐近稳定，当阈值大于1时，传染病平衡点存在唯一，同时得到该传染病平衡点局部指数渐近稳定的充分条件。

The average ADC was lower in the primarily focus of nasopharyngeal non cornification anaplastic carcinoma and metastatic lymph nodes than normal control group.

结果 鼻咽非角化型未分化癌不同中医证型的平均ADC值的差异有显著统计学意义(x2=15.987，P＝0.000)，火毒困结型的平均ADC值明显高于痰浊结聚型、气血凝结型，且差异有显著统计学意义(P＜0.01)；鼻咽非角化型未分化癌原发灶及其转移性颈部、咽后淋巴结的平均ADC值均低于对照组，且差异有显著统计学意义(P＜0.01)。

Record the representative mean, maximum and minimum creepage from the scribe, and note whether or not the maximum is an isolated spot.

记录从刻线处代表性的平均值，最大值，最小值变化，标注最大值是否是一个孤立的点。

But we don't care about Battlegrounds.

但我们并不在乎沙场中的显露。

Ah! don't mention it, the butcher's shop is a horror.

啊！不用提了。提到肉，真是糟透了。