FCM
所属分类：人工智能/神经网络/深度学习
开发工具：Visual C++
文件大小：47KB
下载次数：9
上传日期：2013-04-23 10:12:19
上 传 者：
dujixian255
说明： VC++实现的fcm算法演示程序，FCM算法是一种基于划分的聚类算法，它的思想就是使得被划分到同一簇的对象之间相似度最大，而不同簇之间的相似度最校FCM算法需要两个参数一个是聚类数目C，另一个是参数m。一般来讲C要远远小于聚类样本的总个数，同时要保证C>1。对于m，它是一个控制算法的柔性的参数，如果m过大，则聚类效果会很次，而如果m过小则算法会接近HCM聚类算法。
(Realized VC++ FCM algorithm demo program, FCM algorithm is a clustering algorithm based on partition, the thought of it is making is divided into the greatest similarity between objects of the same cluster, and the similarity between different clusters school FCM algorithm requires two parameters is the number of clusters C, and the other is a parameter m. In general, C is much less than the total number of cluster sample, at the same time to ensure that C> 1. For m, It is a control a flexible algorithm parameters, if m is too large, then the clustering effect will be times, if m is too small, the algorithm will be close to the HCM clustering algorithm.)
文件列表:
data\center.txt (190, 2002-08-10)
data\datain.txt (36775, 2002-08-10)
data\matrix.txt (44384, 2002-08-10)
data\paramOut.txt (24, 2002-08-10)
res\FCM.ico (1078, 2002-07-13)
res\FCM.rc2 (395, 2002-07-13)
~$M聚类算法介绍.doc (162, 2011-05-19)
center.txt (16, 2012-10-20)
FCM.APS (20660, 2012-10-20)
FCM.clw (1060, 2012-10-20)
FCM.cpp (2057, 2012-10-20)
FCM.dsp (4105, 2002-07-13)
FCM.dsw (529, 2002-07-13)
FCM.h (1291, 2002-07-13)
FCM.rc (5216, 2002-07-13)
FCMDlg.cpp (12716, 2012-10-20)
FCMDlg.h (1627, 2002-07-15)
matrix.txt (0, 2012-10-20)
Resource.h (528, 2002-07-13)
StdAfx.cpp (205, 2002-07-13)
StdAfx.h (1056, 2002-07-13)
data (0, 2012-10-20)
Debug (0, 2012-10-20)
res (0, 2012-10-20)
请运行程序，并在程序所在文件夹下面创建文本文件：datain.txt和paramOut.txt，文件的详细说明见下面。程序的运行没有可视的界面，运行完毕之后会在该文件夹下面产生center.txt和matrix.txt两个文件，其中center.txt为聚类的中心，matrix.txt为隶属度矩阵。在运行本程序之前，先参考一下FCM聚类算法介绍.doc
输入文件：
datain.txt
paramOut.txt
其中，datain.txt表示输入的样本数据
paramOut.txt表示输入的聚类参数。
1168
5
4
2
500
.1
其中1168表示样本点的个数，5表示输入样本的维数，4表示要聚类的类别数，2表示m参数，500表示最大的循环次数，0.1表示聚类的精度。
输出文件
center.txt表示聚类的中心，
2.235158,2.216862,2.259246,2.179756,14.487826,
2.011866,1.990132,1.***6863,1.928874,9.807125,
2.159366,2.154501,2.137472,2.051354,21.563583,
1.777569,1.751310,1.846922,1.76***09,2.417483,
一个行表示一个类中心，一个列表示一个维度。
matrix.txt
0.063055,0.129468,0.029966,0.777510,
0.011885,0.030029,0.004839,0.953246,
0.012895,0.035797,0.004966,0.946342,
0.***2582,0.010087,0.005656,0.001675,
0.017419,0.047566,0.006742,0.928273,
0.113058,0.570329,0.031078,0.285535,
0.623048,0.305760,0.040247,0.030945,
0.081656,0.862810,0.014612,0.040923,
0.911263,0.040214,0.040591,0.007932,
0.006618,0.017270,0.002653,0.973458,
0.004695,0.012169,0.001894,0.***1242,
0.004305,0.011272,0.001727,0.***2697,
0.006***3,0.017895,0.002830,0.972292,
一个行表示一个样本属于四个类的隶属度向量，一个列表示属于第几类的隶属度。
本软件源代码公开，如有改动使用请拷贝一份给作者：jakezj@163.com
主页：http://agents.yeah.net
近期下载者：
相关文件：
收藏者：