Persistent contrastive divergence¶

Is a variant of Gibbs sampling, which reuses the same Markov Chain between iterations.

It is an common approach to train Restricted Boltzman Machines.