用matlab实现贝叶斯网络的学习、推理。

This code is to estimate the parameter of model of Mixture Gauss which is popular model used in pattern classification . JAVA 写的估计复合高斯模型中的参数 -This code is to estimate the parameter of model of Jian Gauss which is popular model used in pattern classification. Java was estimated composite Gaussian model parameters

用BP网络实现对图像的边缘进行提取，本程序为演示程序，可以看到BP在应用中的效果-BP network of image edges extraction, the procedure for the demonstration program, we can see BP in the application of the results

关于人工神经网络的实用书籍，适合初学者，以及有一定基础的读者学习-Information on artificial neural networks and practical books, suitable for beginners, as well as readers of a certain basis for learning

人工免疫算法的源代码,经过调试运行,效果还不错,和大家分享!-Artificial immune algorithm source code, after debugging run, the effect is also good to share with you all!

经典的人工智能问题 - iris数据分析问题。通过设计三层bp神经网络对花朵数据进行分类识别，并达到了很好的效果。-classic AI problem- iris data analysis problems. Three-bp through the design of neural networks classify data flower identification, and to achieve good results.

k-means算法 自定义数据样本和类中心 已经调试实现 并且有图清晰表示-k-means algorithm for custom data types of samples and testing center has been and realize that has a clear plan

用VC++实现的BP（反向传播负反馈）神经网络算法-with VC BP (BP negative feedback) neural network algorithm

基于ICA的信号分离程序，是一个很好的信号处理方面的程序-based ICA signal separation procedure is a very good signal processing procedures

不确定规划 遗传算法 Nonlinear Programming算法程序- Indefinite plan heredity algorithm Nonlinear Programming algorithm procedure

自组织系统Kohonen网络模型。对于Kohonen神经网络，竞争是这样进行的：对于“赢”的那个神经元c，在其周围Nc的区域内神经元在不同程度上得到兴奋，而在Nc以外的神经元都被抑制。网络的学习过程就是网络的连接权根据训练样本进行自适应、自组织的过程，经过一定次数的训练以后，网络能够把拓扑意义下相似的输入样本映射到相近的输出节点上。网络能够实现从输入到输出的非线性降维映射结构：它是受视网膜皮层的生物功能的启发而提出的。~..~-Kohonen network model. For Kohonen neural network, competition is this : For the "winner" of neurons c, in its switching around the region neurons in varying degrees, to be excited, and the switching outside the neurons were inhibited. Network learning is a process in the network connecting the right under the training samples for adaptive, self-organizing process, after a certain number of training, network topology can sense similar to the mapping of the input samples similar to the output nodes. Network can be achieved from input to output of nonlinear reduced-dimensional mapping structure : it is subject to retinal cortex of the biological function inspired by. ~ ~ ..