I think that the equation on line 50 in the BP and JT notes is missing a factor of $$\phi(x_i)$$. Since local single node potentials are included in the messages sent from $$x_k$$ to $$x_j$$ in the preceding equation, shouldn't we consider the local single node potential at node $$x_i$$ when computing its marginal?
I think that the equation on line 50 in the BP and JT notes is missing a factor of $$\phi(x_i)$$. Since local single node potentials are included in the messages sent from $$x_k$$ to $$x_j$$ in the preceding equation, shouldn't we consider the local single node potential at node $$x_i$$ when computing its marginal?