1 paper
Marco Zaffalon, Alessandro Antonucci
We characterise the likelihood function computed from a Bayesian network with latent variables as root nodes. We show that the marginal distribution over the remaining, manifest, v…