java系列归因于反

利用SOM（Self-Organization Map）数据聚类方法，包含点数据集的坐标，并以TChart分组为例进行说明。

基于MATLAB的神经网络非线性系统辨识软件包.-MATLAB-based nonlinear neural network system identification package.

本人编写的GramSchmidt神经网络算法，和传统的神经网络算法比较，该算法能自动产生神经元，节约时间，速度快！ 本人还加了自动产生分类的方法和适合各种函数的转换，该zip包括两个文件，一个是GramSchmidt正交分解，是个子函数，主函数是orth_GramSchmidt，放在同一目录下，就可运行，可以对benchmark的逼近和分类问题提供解决方案-I prepared GramSchmidt neural network algorithm, and the traditional neural network algorithm, the algorithm can automatically generate neurons, saving time, fast! I also added a method of automatically generated classification and conversion for a variety of functions, the zip includes two documents, one is GramSchmidt orthogonal decomposition is a sub-function, the main function is orth_GramSchmidt, on the same directory, you can run can benchmark approximation and classification problems to provide solutions

最新开发的小波融合算法实现VC程序。非常好用。-latest development of the wavelet algorithm VC fusion procedures. Very handy.

国外网站上的svm源码，希望对大家有帮助-abroad on the website svm source, and I hope to be helpful

该程序是基于BP网络的PID控制系统。系统由两部分构成：经典的PID控制器和神经网络。-that the procedure was based on BP network PID control system. System consists of two parts : the classic PID controller and neural networks.

eview 编写的时间序列预报程序 可以用于金融预报，等-eview the preparation of time-series forecasting procedures can be used for financial forecasting, etc.

pre-fixSpan算法，该算法是C++语言实现的，主要用模板实现prefixspan的投影数据库架构及模式生长功能

仿真1：首先把网络温度参数T固定在100，按工作规则共进行1000次状态更新,把这1000次状态转移中网络中的各个状态出现的次数Si（i＝1，2，…，16）记录下来 按下式计算各个状态出现的实际频率： Pi=Si/∑i=1,NSi=Si/M 同时按照Bo1tzmann分布计算网络各个状态出现概率的理论值： Q（Ei）=（1/Z）exp（-Ei/T） 仿真2：实施降温方案，重新计算 采用快速降温方案：T（t）= T0/（1+t） T从1000降到0.01，按工作规则更新网络状态 当T=0.01时结束降温，再让T保持在0.01进行1000次状态转移，比较两种概率-a simulation : First of all network parameters temperature T fixed at 100 and, according to the rules for a total of 1000 to update the state, this state of the 1000 network transfer of all states for the number of Si (i = 1, 2, ..., 16) all recorded determined by the formula state-of the actual frequency : Pi = Si/i = 1, NSi = Si/M in accordance with Bo1tzmann distributed computing network of states all probability the theoretical value : Q (Ei) = (1/Z) exp (- Ei/T) Simulation 2 : implementation of cooling, re-using rapid cooling programs : T (t) = T0/(1 t) T dropped to 0.01 from 1000 and, according to the rules updated network state when T = 0.01 at the end of cooling, let T at 0.01 for the 1000 state tran

炉温模糊控制的一般性程序，可以修改各参数从而更改模糊控制表-Temperature control of the general procedures can be amended so as to change the parameters of fuzzy control Table