局部网络

Aiming at the problems of over-fitting and generalization for neural networks, the locally most entropic colored Gaussian joint input-output probability density function estimate is studied, and a new method by means of Chebyshev inequality is proposed to self-revise respectively according to every cluster.

针对神经网络的过拟合和泛化能力差的问题，研究了样本数据的输入输出混合概率密度函数的局部最大熵密度估计，提出了运用Chebyshev不等式的样本参数按类分批自校正方法，以此估计拉伸样本集，得到新的随机扩充训练集。

Experimental result indicates, the optimized BP neural network's constringency speed is faster, and the problem that BP algorithm is easy to be trapped into local optimization has been solved.

试验结果表明，用遗传算法优化BP神经网络的连接权值后收敛速度快，并有效的解决了BP算法容易陷入局部最优的问题。

Based on these, we formulate the correspondence problem as a constrained optimization problem, and use Hopfield neural-networks to solve it.

在Hopfield网络的能量函数的设计中，综合考虑了特征点的预测位置及运动轨迹在局部所呈现的平滑性。

Based on the research work of basic theories including phase space reconstruction,embedding theorem,correlation dimension,local dynamics,Lyapunov exponents,surrogate data etc,based on the research work of general methods such as principal component analysis,correlation dimension GP algorithm,false neighbors method,nonlinear time series prediction,local prediction, adaptive prediction,neural network model,support vector machines regression model, prediction power,nonlinear detection,coarse-graining methodology,conditional entropy and so on,the framework of nonlinear time series analysis are constructed.

在对包括相空间重构、嵌入定理、关联维数、局部动力学、Lyapunov指数、替代数据(来源：A97BC论文网www.abclunwen.com)、等基本理论与其物理意义的研究和讨论基础上；在对包括主分量分析、关联维数GP算法、伪邻近点法、非线性时间序列预测、局域预测、自适应预测、神经网络模型、支持向量回归模型、预测效果、非线性检测、粗粒化方法、条件熵等非线性时间序列一般分析方法的原理和算法研究基础上；构建了新的非线性时间序列分析的理论体系，归纳总结了非线性时间序列分析的基本问题和主要研究方面。2。

It can enhance generalization ability of neural network and distill the rules automatically by using the new method.

为了避免单纯使用神经网络易陷入局部极小的缺点，本文还提出了一种融合遗传算法？？

The experimental results show that temperature error and non-linearity of silicon piezoresistive pressure sensor can be reduced markedly.

本文针对其缺点，提出了以正交的局部函数——小波函数作为激励函数的小波神经网络对硅压阻式压力传感器的温度误差进行补偿。

Under the deeply research and a large mount of experiments, the paper put forwards a series of methods to solve local extreme value problem.

本文通过深入研究与大量的神经网络应用，探讨出一系列解决局部极值解问题的方法，实验结果表明，这些方法是有效的，有一定的理论和应用价值。

The five characteristics of the system are exploiting the ordered decomposition of problems in the structure of networks; using the Hasse graph to map the transfer network to the field of real numbers homomorphically; distinguishing the production rules as strong rules and weak rules; using the method of scheduling among stations to balance the search by weak rules in every station; searching in the best-first way both in the large and in the small, but with the different aims respectively, which can be regarded as a BF′-BF strategy in which the macro guides the micro.

该系统有五个特点：利用了一种网络结构的有序分解；采用Hasse图将中转站网络同态地映射到实数域上；将产生式规则区分为强规则和弱规则；采用站间调度的方法系统地平衡了各站弱规则的搜索；全局和局部都为最佳优搜索，只是目标指向不一样，是一种宏观引导微观的BF′-BF混合策略。

It is import of choosing a basis function in the theory and application of the locally approximate neural networks.

在局部逼近神经网络的理论研究和实际应用中，基函数的选取是非常重要的。

Substituting other faster algorithm of NN for the BP algorithm in it, we get a new non-line homotopic synthetic learning algorithm, which increases the speed of convergence and improves the ability of avoiding local minimum.

在此基础上，进一步提出用其它改进的神经网络快速算法代替其中的BP算法，从而形成新的非线性同伦综合学习算法，即保持了快速收敛的特点，又较大地提高其避免陷入局部极小的能力，收到很好的训练效果。

On the other hand, the more important thing is because the urban housing is a kind of heterogeneity products.

另一方面，更重要的是由于城市住房是一种异质性产品。

Climate histogram is the fall that collects place measure calm value, cent serves as cross axle for a few equal interval, the area that the frequency that the value appears according to place is accumulated and becomes will be determined inside each interval, discharge the graph that rise with post, also be called histogram.

气候直方图是将所收集的降水量测定值，分为几个相等的区间作为横轴，并将各区间内所测定值依所出现的次数累积而成的面积，用柱子排起来的图形，也叫做柱状图。