人工神经网络BP算法 1、动态改变学习速率 2、加入动量项 3、运用了Matcom4.5的矩阵运算库（可免费下载,头文件matlib.h)， 方...

神经网络源码可应用于遥感图像分类，包括bp、kohonen等方法-neural network source can be used in remote sensing image classification, including bp, Kohonen methods

改进遗传算法程序-improved genetic algorithm program

Adaptive Resonance Theory (ART) Neural Network

这是好用的源代码-This is the source code handy

一个使用Java实现的人工智能的图搜索策略程序――酒九宫图问题 人工智能－图搜索策略－演示程序－使用说明 程序简介 在人工智能领域，对隐式图的搜索是求解问题的一种基本方法，本程序通过使用多种图搜索策略，对 重排九宫问题进行求解，演示了“广度优先搜索”、“深度优先搜索”、“有界深度优先搜索”、 “最好优先搜索”和“局部择优搜索”五种基本的图图搜索策略。 整个程序使用了动画技术，界面设计美观友好，使用方便。-use a Java artificial intelligence plans Search Strategy-- Liquor JiuGongTu problems of artificial intelligence-map search strategy- Demonstration Program-use guide in the field of artificial intelligence, the implicit search of the plan to solve the problem is one of the basic methods, the procedures through the use of multiple map search strategy, the rearrangement JiuGongTu problem solving, presentation of the "breadth- first search," "depth-first search", "bounded depth-first search", "best-first search" and "partial selection of the best search" five basic map search strategy. The whole process of using animation, interface design aesthetic friendly, easy to use.

遗传算法解决双变量的函数最优化问题，有按钮的界面，用bc所编，生动模拟遗传进化过程-genetic algorithm to solve the two- variable optimization function, the button interface, using bc prepared by the vivid simulation of the process of genetic evolution

MYCIN不确定推理机的C++骨架程序，运行高效率。-MYCIN uncertainty reasoning machine C++ skeleton program, run efficient.

模式识别中关于C均值的聚类算法的程序,输入点进行聚类.-pattern recognition on the C-means clustering algorithm procedures input cluster.

一个分词程序，c代码，有很详细的注释，便于阅读-A segmentation procedure, c code, have very detailed notes, easy-to-read

统计学习理论是建立在一套较坚实的理论基础之上的，为解决有限样本学习问题提供了 一个统一的框架。它能将很多现有方法纳入其中，有望帮助解决许多原来难以解决的问题（比 如神经网络结构及参数选择问题）；同时。在这一理论基础上发展了一种新的通用学习方法： 支持向量机（Support Vector Machine，SVM）。它己表现出很多优秀的性能，并已经成为当 今机器学习领域的研究热点。-Statistical learning theory is based on a more solid theoretical basis, in order to solve the learning problems of limited samples provides a unified framework. It will be incorporated into many existing methods, which could help solve many of the original difficult to resolve problems (such as neural network structure and parameters selection) the same time. Developed on the basis of this theory, a new universal learning methods: support vector machine (Support Vector Machine, SVM). It has been shown a lot of excellent performance and has become the hotspot of machine learning research in the field.