网上的一个BP算法实现数字识别的demo，希望能对其开发者有帮助...

支持向量机Windows Svm-Support Vector Machines of Windows Svm

一个很好的中国象棋源代码 本人机对弈程序采用了多种搜索算法. 本程序还具有悔棋,还原功能,还可以记录走法.-a good source of Chinese chess game machine code, I use a variety of procedures for search algorithm. The procedure also has to repent Mullick, recovery capabilities, but also can be recorded in law.

一个非常玄的选择颜色的例子，颜色选择通过三维立体显示，不可多得-A very mysterious example of the choice of color, color selection by three-dimensional stereoscopic display, a rare

machine pulse control

　人工神经网络(Aartificial Neural Network，下简称ANN)是模拟生物神经元的结构而提出的一种信息处理方法。早在1943年，已由心理学家Warren S.Mcculloch和数学家Walth H.Pitts提出神经元数学模型，后被冷落了一段时间，80年代又迅猛兴起［1］。ANN之所以受到人们的普遍关注，是由于它具有本质的非线形特征、并行处理能力、强鲁棒性以及自组织自学习的能力。其中研究得最为成熟的是误差的反传模型算法(BP算法，Back Propagation)，它的网络结构及算法直观、简单，在工业领域中应用较多。-Artificial Neural Networks (Aartificial Neural Network, under the short title of ANN) is a simulation of the structure of biological neurons and the information put forward a method of treatment. At early 1943, by Warren S. Mcculloch psychologists and mathematicians Walth H. Pitts proposed mathematical model of neurons, was left out in the cold for some time, and the rapid rise of the 80" s [1]. ANN because the general concern of the people is the essence of it is because of the non-linear characteristics of parallel processing capabilities, strong robustness and self-organizing capacity of self-learning. Research one of the most sophisticated are the error-propagation model of the algorithm (BP algorithm, Back Pro

神经网络OOP类库，含有BP网络等的训练，很容易对其进行扩展，-neural network OOP libraries, containing BP network of training, it is easy to extend them

不错的自组织神经网络算法，程序有注释。能运行，效果还行。-good self-organizing neural network algorithm, the procedure Notes. Can run, the results were OK.

vb的波尔兹曼机（含模拟退火算法），有需要的可参考一下 -vb the Boltzmann machine (with simulated annealing algorithm), the need of a reference

书籍“Regularization tools for training large feed-forward neural networks using Automatic Differentiation”的源码文件-Books

通过C4.5 的实现可以进行构建决策树 来进行有效的分类-Through the realization of C4.5 decision tree can be constructed to carry out an effective classification