GSoC Week 5

The week 5 began with a discussion with whether we should deprecate params. I fixed some bugs in checking functions, random number generator and one of covariance updating methods. In the following days, I completed the main functions of GaussianMixutre and all test cases, except AIC, BIC and sampling functions. The tests are some kind of challenging, sine the current implementation in the master branch contains very old test cases imported from Weiss’s implementation which is never got improved. I simplified the test cases, and wrote more tests that are not covered by the current implementation, such as covariance estimation, ground truth parameter prediction, and other user-friendly warnings and errors.

Next week, I will begin to code BayesianGaussianMixture.

GSoC Final Project Report

GSoC is approaching its end. I am very glad to have such great experience this summer.I explored the classical machine learning models, G...… Continue reading

Progress Report 3

Published on August 07, 2015

GSoC Week 8, 9 and Progress Report 2

Published on July 27, 2015