1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
|
Last Change: Fri Aug 04 07:00 PM 2006 J
Version 0.4.2
pyem is a python module build upon numpy and scipy
(see http://www.scipy.org/) for learning mixtures models
using Expectation Maximization. For now, only Gaussian
Mixture Models are implemented. Included features:
* computation of Gaussian pdf for multi-variate Gaussian
random vectors (spherical, diagonal and full covariance matrices)
* Sampling of Gaussian Mixtures Models
* Confidence ellipsoides with probability (fixed level of
0.39 for now)
* Classic EM for Gaussian Mixture Models
* K-mean based and random initialization for EM available
Has been tested on the following platforms:
* Ubuntu dapper, bi Xeon 3.2 Ghz, 2 Go RAM
python 2.4 + pyrex, numpy 1.0.b2SVN + scipy 0.5.1SVN, uses atlas3-sse2
* Ubuntu dapper, pentium M 1.2 ghz,. 512 Mo Ram
python 2.4 + pyrex, numpy 1.0.b2SVN + scipy 0.5.1SVN, uses atlas3-sse2
* Ubuntu dapper, minimac (ppc G4 1.42 Ghz, 1Gb RAM)
python 2.4 + pyrex, numpy 1.0.b2SVN + scipy 0.5.1SVN, uses atlas3-sse2
|
