Learning Graphical Model Parameters with Approximate Marginal Inference

Justin Domke

Likelihood based-learning of graphical models faces challenges of computational-complexity and robustness to model mis-specification. This paper studies methods that fit parameters…