|
What is the intuition behind message passing algorithms? I understand the graph is converted to a clique tree, and then messages are passed between cliques( which propagate belief of each nodes). But how comes when it converges, each belief represent the marginal probability of variables? |
