tsp问题用神经网络方法解决

一个用于聚类的工具箱，内有主元分析、模糊等技术的Matlab源代码和应用实例程序Demo。-One for clustering toolbox, with principal component analysis, fuzzy techniques, such as Matlab source code and application procedures for Demo.

Eigenface and Fisherface Performance Across Pose

仿真1：首先把网络温度参数T固定在100，按工作规则共进行1000次状态更新,把这1000次状态转移中网络中的各个状态出现的次数Si（i＝1，2，…，16）记录下来 按下式计算各个状态出现的实际频率： Pi=Si/∑i=1,NSi=Si/M 同时按照Bo1tzmann分布计算网络各个状态出现概率的理论值： Q（Ei）=（1/Z）exp（-Ei/T） 仿真2：实施降温方案，重新计算 采用快速降温方案：T（t）= T0/（1+t） T从1000降到0.01，按工作规则更新网络状态 当T=0.01时结束降温，再让T保持在0.01进行1000次状态转移，比较两种概率-a simulation : First of all network parameters temperature T fixed at 100 and, according to the rules for a total of 1000 to update the state, this state of the 1000 network transfer of all states for the number of Si (i = 1, 2, ..., 16) all recorded determined by the formula state-of the actual frequency : Pi = Si/i = 1, NSi = Si/M in accordance with Bo1tzmann distributed computing network of states all probability the theoretical value : Q (Ei) = (1/Z) exp (- Ei/T) Simulation 2 : implementation of cooling, re-using rapid cooling programs : T (t) = T0/(1 t) T dropped to 0.01 from 1000 and, according to the rules updated network state when T = 0.01 at the end of cooling, let T at 0.01 for the 1000 state tran

一个用c++builder实现的神经网络算法的程序-c builder with a realization of the neural network algorithm for the procedure

遗忘因子法递推算法用于估计数学模型的参数,精确度较高!-Forgetting factor recursive algorithm method used to estimate the parameters of the mathematical model, high accuracy!

这个是matlab的BP 神经网络-This is matlab" s BP neural network. . . . . . . . . . . . .

变色龙层次聚类算法，利用动态模型的层次聚类算法-chameleon hierarchical clustering algorithm, using a dynamic model of hierarchical clustering algorithm

这个软件可以对输入的象素矩阵进行识别。是根据《Neural Network Design》（Martin T. Hagan等）中的关于有监督的Heb学习编写的。使用JBuilder开发的界面，使用JCreator开发的核心。-the software can input pixel matrix identification. Based on "Neural Network Design" (Martin T. Hagan, etc.) on the supervision of a study prepared by the Heb. Using JBuilder development of the interface, use JCreator development of the core.

基于粒子滤波原理，选取颜色特征对目标进行描述，实现视频目标的实时跟踪，效果较好。-Principle-based particle filter, select the color characteristics of the target description, realize the goal of real-time video tracking, better.

KNN和BP的算法实现。输入一系列的样本值，进行训练和分类。源码是实现了两个分类器，已经调试通过-KNN and BP algorithm. Enter a series of sample values, for training and classification. Source is to achieve the two classifiers have been debugging through