小神经

Methods: Study the newly found cell with the weak silver carbonate staining method of del Rio-Hortega. Using transmission electron microscopy, compare the newly found cell with the atypical cell found by Marc Lenoir and Philippe Vago, and extend young SD rat model to adult SD rat, young and adult Wistar rat,and extend drug of Neomycin to Amikacin to study the universality.

采用经典的小胶质细胞特殊染色方法——银染法(Hortega氏碳酸银法)对大鼠耳蜗药物损伤后出现的新细胞进行研究；通过透射电镜与Marc Lenoir和Philippe Vago实验动物模型中的"非典型细胞"进行了比较与扩展研究；运用逆转录PCR技术检测耳蜗基底膜上神经干细胞标志物nestin的表达情况，对新细胞的来源及其与神经干细胞或nestin的前体细胞的关系作进一步研究。

There are five important soft sensor models of moisture have been established in this paper: the moisture model of The Recursive Least Square based on Lethe gene, the moisture model of Principal Component Regression , the moisture model of Partial Least Square, the moisture model of BP Nerve Network, the moisture model based on PLS—BP. All of these soft sensor models of moisture are systematically analysed by using Compound Correlative Coefficient. The factors which influence control of moisture are discussed on detail in this paper.

建立了五个重要的水分软测量模型，即渐消记忆递推最小二乘法水分模型；主成分分析法水分模型；偏最小二乘法水分模型；BP 神经网络水分模型；偏最小二乘BP 网络水分模型；对以上水分软测量模型用统一的评价指标复相关系数对它们的优劣进行了详细的分析，选出最适合水分建模的建模方法，同时对影响水分的各个因素进行了深入的分析和探讨。

Aiming at the drawbacks of traditional BP neural network, such as converging slowly and tending to get into the local minimizer, the particle swarm algorithm is introduced into the training of neural networks.

此外针对传统的BP神经网络存在收敛速度慢以及容易陷入局部极小值的缺点，引入粒子群算法对神经网络的学习训练进行优化。

The results showed that IL-1 receptor-like immunoreactive neurons existed in both nodose ganglion and vagal paraganglion in normal rat. In nodose ganglion most of cells were small or medium in size, whereas almost all cells in vagal paraganglion were positive.

结果表明，正常大鼠结状神经节和迷走旁节中均存在IL-1Ⅰ型受体样免疫反应阳性的神经元，结状神经节中阳性神经元以中、小细胞为主；迷走旁节中几乎所有的细胞均呈IL-1Ⅰ型受体样免疫阳性。

Orthogonal least square is a widely used approach to training networks, so it also can be applied to wavelet networks.

正交最小二乘法是一种常用的前向神经网络训练方法，因此可以用正交最小二乘法训练小波网络。

This thesis discussed the recursive least square algorithm, the linear least square algorithm and recursive orthogonal least square algorithm for the RBPNNs.

文中讨论了递推最小二乘算法、线性最小二乘法和递推正交最小二乘算法用于径向基概率神经网络的训练问题。

This network is capable of clustering the QRS waves in the automatic analysis of ECG signals.

介绍一个由小波变换和神经网络结合的自组织联想神经网络模型，它可用在心电信号的自动分析中对QRS波进行聚类。

According to the input of the sample set, a function set of multiresolution wavelet is obtained and the output vectors of each node are also computed.

总结了多层前向神经网络的结构设计方法，然后提出了一种新的小波神经网络构造算法。

On the other hand, this dissertation analyzes the difference between wavelet function and sigmoid function.

在网络结构相同的情况下，连续小波神经网络的预测精度能够超过BP神经网络。

There are some disadvantages in BP algorithm, such as low rate of convergence, easily falling into local minimum point and weak global search capability. In order to settle these problems, a genetic algorithm was used to train BP neural network to replace classical learning algorithms. An evolutionary neural network learning algorithm was founded.

针对BP算法存在收敛速度慢、容易陷入局部极小值及全局搜索能力弱等缺陷，采用遗传算法训练BP神经网络，取代了一些传统的学习算法，设计了基于进化神经网络的学习算法。

The split between the two groups can hardly be papered over.

这两个团体间的分歧难以掩饰。

This approach not only encourages a greater number of responses, but minimizes the likelihood of stale groupthink.

这种做法不仅鼓励了更多的反应，而且减少跟风的可能性。