Clustering

Efficient implementation of the clustering using the Expectation Maximization (EM) and K-means algorithms

Requirements

• Python 2.7 or Python 3.3
• scipy
• numpy

Usage

To apply the EM algorithm on a dateset one needs to call:

$  L = gmm.EM(max_iter=max_iter, tol = tol)

To apply the K-means algorithm on a dataset one needs to call:

$ D = gmm.k_means(max_iter=max_iter, tol=tol)

To sample from an already trained GMM model, one needs to call:

$ Y = gmm.sample(N=N)

Several examples can be found within the main.py file.

Sample results

[1] Results of the EM algorithm over iterations

Sample data

[1] Unlabelled data

[2] Labelled data

Results of the EM algorithm

[1] Final result of the EM algorithm

[2] Final result of the EM algorithm (covariance matrices)

[3] EM cost function over iterations

Results of the K-means algorithm

[1] Final result of the K-means algorithm

[2] K-means cost function over iterations

Sampling results

[1] Sampling results from already trained GMM model