Overview
	• Clustering with K-means and a proof of convergence that uses energies. • Clustering with a mixture of Gaussians and a proof of convergence that uses free energies • The MDL view of clustering and the bits-back argument • The MDL justification for incorrect inference.