Towards a unified recurrent neural network theory: The uniformly pseudo-projection-anti-monotone net

In the past decades, various neural network models have been developed for modeling the behavior of human brain or performing problem-solving through simulating the behavior of human brain. The recurrent neural networks are the type of neural networks to model or simulate associative memory behavior of human being. A recurrent neural network (RNN) can be generally formalized as a dynamic system associated with two fundamental operators: one is the nonlinear activation operator deduced from the input-output properties of the involved neurons, and the other is the synaptic connections (a matrix) among the neurons. Through carefully examining properties of various activation functions used, we introduce a novel type of monotone operators, the uniformly pseudo-projectionanti-monotone (UPPAM) operators, to unify the various RNN models appeared in the literature. We develop a unified encoding and stability theory for the UPPAM network model when the time is discrete. The established model and theory not only unify but also jointly generalize the most known results of RNNs. The approach has lunched a visible step towards establishment of a unified mathematical theory of recurrent neural networks.     

不等式组与变分不等式的极大熵函数方法

利用极大熵函数方法将不等式组及变分不等式的求解问题转化为近似可微优化问题，给出了不等式组及变分不等式问题近似解的可微优化方法，得到了不等式组和变分不等式问题的解集合的示性函数．     

复正定矩阵的Minkowski不等式

建立了复正定矩阵的几个行列式不等式,将正定Ｈｅｒｍｉｔｅ阵的Ｍｉｎｋｏｗｓｋｉ不等式、 Ｏｓｔｒｏｗｓｋｉ－Ｔａｕｓｓｋｙ不等式推广到了复正定矩阵上,推广改进了一些文献的结果．     

初始点任意的一个非线性优化的广义梯度投影法

广义投影算法的优点是避免转轴运算。它成功地给出了线性约束问题、初始点任意的只带非线性不等式约束问题，以及利用辅助规划来处理带等式与不等式约束问题的算法．后者完满地解决了投影算法对于非线性等式约束问题的处理，但要求满足不等式约束的初始点．本文据此利用广义投影与罚函数技巧给出了一个初始点任意的等式与不等式约束问题的算法，省去了求初始解的计算，并保持了上述方法的优点，证明了算法的全局收敛性     

一类与凸函数有关的不等式的概率证法

通过构造随机变量的方法,并结合凸函数的性质,证明了数学分析中的一些著名不等式,证明方法简明、独特.     

一个有趣的不等式

利用凸函数的性质,证明了一个不等式     

关于一类递推数列极限的求法的注解

证明了一类递推数列xn 1=(xn b)/(xn c)(ac≠b,n=1,2,……)的收敛性,并给出这类数列的一些性质.本文所得结果推广和包含了文[1]的相应结果.     

Convergence of BP Algorithm for Training MLP with Linear Output

The capability of multilayer perceptrons(MLPs)for approximating continuous functions with arbitrary accuracy has been demonstrated in the past decades.Back propagation(BP)algorithm is the most popular learning algorithm for training of MLPs.In this paper,a simple iteration formula is used to select the leaming rate for each cycle of training procedure,and a convergence result is presented for the BP algo- rithm for training MLP with a hidden layer and a linear output unit.The monotonicity of the error function is also guaranteed during the training iteration.     

Rn中的广义逆Bonnesen型不等式

等周问题在积分几何中具有举足轻重的地位.该文主要研究Rⁿ中等周不等式的逆形式,即广义逆Bonnesen型不等式.该文获得了Rⁿ中几个新广义等周亏格上界的结果,作为推论,得到了更一般的平面上的逆Bonnesen型不等式;最后给出其中三个上界结果之间的最佳估计.     

分辨率优化的混合WENO格式

为提高有限差分格式的分辨率,利用傅里叶分析对WENO格式进行色散及耗散优化,并给出优化的线性权重.用优化后的WENO格式与保单调格式（MP）进行加权混合,得到新的加权混合WENO格式（H-WENO）.通过一维激波管问题、Shu-Osher问题及二维双Mach反射问题及R-T不稳定性问题对格式进行数值测试.结果显示,新格式具有强健的激波捕捉能力和对小尺度波结构的高分辨率,与原WENO格式相比改进明显.