BP算法

非线性程序之模式识别与控制系统设计.-nonlinear pattern recognition procedures and control system design.

本程序是使用一种遗传算法解决m个工件分配给m架机床的效益最优化问题，-this program is to use a genetic algorithm to solve m m allocated to the workpiece-machine efficiency optimization,

基于倒谱特征的带噪语音端点检测.rar格式为ｖｉｐ-based Cepstral Noisy voice endpoint detection. Rar format vip

用于分类的一个工具箱，内含支持向量机、神经网络、主元分析、多变量样条等方法的matlab源代码，还有用户手册。-A tool for classification, including support vector machines, neural networks, principal component analysis, multivariate methods such as spline matlab source code, as well as user manuals.

加了注释的c代码（complexNetwork）源代码，下载后就可以直接使用-Added a comment c code (complexNetwork) source code, downloads can be used directly after

模式识别中的寻路算法：A星算法的实现及简单应用-Pattern Recognition routing algorithm: A Star algorithm and a simple application

把libsvm包在一起，只要能读入矩阵转换成所需的格式libsvm就可以运行了

神经网络C程序的第8章-neural network C procedures for the Chapter 8

用Java实现的多层BP神经网络，用于颜色和字符识别-using Java multi-storey BP neural network, for the color and character recognition

由于K－均值聚类算法局部最优的特点，而模拟退火算法理论上具有全局最优的特点。因此，用模拟退火算法对聚类进行了改进。20组聚类仿真表明，平均每次对K结果值改进8次左右，效果显著。下一步工作：实际上在高温区随机生成邻域是个组合爆炸问题（见本人上载软件‘k－均值聚类算法’所述），高温跳出局部解的概率几乎为0，因此正考虑采用凸包约束进行模拟聚类，相关工作正在进行。很快将奉献给各位朋友。-as K-means clustering algorithm for optimal local characteristics, and simulated annealing algorithm theory with the characteristics of the global optimum. Thus, simulated annealing algorithm for clustering improvements. Cluster Group of 20 simulations show that the average value of K results improved about eight times, the results are obvious. The next step : In fact, in high temperature generated random neighborhood is a combination of explosives (see my software on the "k-means clustering algorithm" mentioned above), high-temperature solution of partial out almost zero probability, it is considering the use of convex hull bound for simulation cluster, the work under way . Soon dedication to the ladies.