Mini-Projects for MPhil in Scientific Computing, University of Cambridge
Implement and compare different methods for clustering:
- K-Means
- Mixture of Gaussians
- Dirichlet Process Mixture Model
The DPM in particular is very exciting. Implementation uses a Gaussian Likelihood and a Normal-Inverse-Wishart prior. Sampling is done using "Algorithm 2" in Neal(2000).
Written in C++ using the Armadillo library. Prototyped in Matlab.
Tested on various data sets