C-Means

This is an implementation of fuzzy c-means (a.k.a k-means) based on this paper (in Japanese). c_means.m is a generic fuzzy c-means source code. By inputting a fuzzy max function and its gradient function, a specific fuzzy c-means algorithm is performed. The way of usage may be found in plot_clustering.m, which shows the result of c_means.m for data/norm5.mat data set.

Execution

In plot_clustering.m, set

max_func.val = @max_val;
max_func.grad = @max_grad;

and run

>> plot_clustering

then, we get the result of c-means as follows.

If you want to do the fuzzy c-means with the entropy regularization, set

nu = 10
max_func.val = @(X) entro_val(X,nu);
max_func.grad = @(X) entro_grad(X,nu);

ane run

>> plot_clustering

then, we get the result of c-means as follows.

Examples

Bezdek's fuzzy c-means for parameter 2 and 20

Fuzzy c-means with the entropy regularization for parameter 10 and 30

Fuzzy c-means with the quadratic regularization for parameter 20 and 60