Finding the conditional distributions of a Gaussian Mixture Model
Gaussian Mixture Models are a classic clustering technique, that can easily be generalised to, for example, semi-supervised learning. Sometimes we need to compute closed-form expressions for the conditional distributions. Finding this a bit more difficult that expected, these notes and code were created as a way of making sure that I'd done it properly...
Feedback welcome - email is p.l.green@liverpool.ac.uk