This repository contains code written during my Ph.D thesis on clustering models for Gaussian mixture models.
Note: this code is provided 'as-is' and need refactoring to be used. It relies heavily on Numba, comment @jit decorator for testing.
Several portions of this code come from different sources, some of which I did not referenced properly, nevertheless, I thank the authors.
the models present in this repository are:
- Graphical lasso on Gaussian Mixture models
- Square root lasso on Gaussian Mixture models
- Estimation of number of clusters by regularization of posterior probabilities
- Estimation of number of clusters by penalizing the weight vector.
I also implemented
- Projection on the probability simplex from http://arxiv.org/pdf/1309.1541.pdf in slope_implementation.py