BP算法是由学习过程由信号的正向传播与误差的反向传播两个过程组成。由于多层前馈网络的训练经常采用误差反向传播算法，人们也常把将多层前馈网络直接称为BP网络。

最全的复杂网络的MATLAB工具箱，内含大量源代码，直接可用，找了很久才找到的的，觉得不错记得分享应用心得哦！

该工具箱提供了实现curvelet的两种快速算法，USFFT和wrapping

Versions of the toolbox up to 1.9 require MATLAB® Version 7.4 (R2007a) or later. Version 2 of the toolbox requires MATLAB® Version 7.14 (R2012a) or later.

创建你的第一个贝叶斯网络 手工创建一个模型 从一个文件加载一个模型 使用 GUI 创建一个模型 推断 处理边缘分布 处理联合分布 虚拟证据 最或然率解释 条件概率分布 列表（多项式）节点 Noisy-or 节点 其它（噪音）确定性节点 Softmax（多项式 分对数）节点 神经网络节点 根节点 高斯节点 广义线性模型节点 分类 / 回归树节点 其它连续分布 CPD 类型摘要 模型举例 高斯混合模型 PCA、ICA等 专家系统的混合 专家系统的分等级混合 QMR 条件高斯模型 其它混合模型 参数学习 从一个文件里加载数据 从完整的数据中进行最大似然参数估计 先验参数 从完整的数据中（连续）更新贝

Mateda2.0一个非常好用的分布估计算法matlab工具箱，For a preliminary explanation of Mateda2.0 see the file Mateda2.0-UserGuide.pdf in this directory. General documentation about the programs is available in the /doc directory or from: http://www.sc.ehu.es/ccwbayes/members/rsantana/software/matlab/MATEDA.html MATEDA-2.0 employs the Matlab Bayes Net (BNT) toolbox (Murphy:2001) and the BNT structure learning package (Leray_and_Francois:2004). These programs, which are freely available from the authors website (they can be respectively downloaded from http://people.cs.ubc.ca/~murphyk/Software/BNT/bnt.html and http://banquiseasi.insa-rouen.fr/projects/bnt-slp/), should be installed previously to the MATEDA-2.0 installation. Some of the MATEDA-2.0 routines also employs the MATLAB statistical toolbox and the affinity propagation clustering algorithm (Frey_and_Dueck:2007) (the Matlab implementation of affinity propagation is available from http://www.psi.toronto.edu/affinitypropagation/). After installing the BNT and BNT structure learning tools: 1) Set the path to the current BNT structure learning tool directory. This is done by modifying file add_SLP.m. 2) Unpack the file IntEDA.zip and copy the files to a directory named MATEDA. 3) Edit file InitEnvironment.m updating the paths path_MATEDA, path_FullBNT and path_BNT_SLP. 4) Set the current Matlab directory to the MATEDA directory. 5) Execute program InitEnvironments.m. Several warnings but no error should appear. The folder ScriptsMateda contains several examples of EDAs implementations. The file Mateda2.0-UserGuide.pdf contains a detailed explanation of how to use the programs. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. Last version 2/04/2009. Roberto Santana (roberto.santana@ehu.es)

论文，基于Matlab工具箱的电力系统小干扰稳定性分析，可以了把

Kalman滤波Matlab工具箱，包含有关Kalman滤波的一些函数，包含一个简单的目标跟踪仿真程序

最新版，最小二乘支持向量机（2010年9月更新）matlab工具箱（内附说明文档（英文版），文档中附有例子）

matlab里可用的nurbs工具箱，可以用来进行生成nurbs曲线曲面。