1: \begin{abstract}

2: Many Bayesian statistical inference problems come down to computing a maximum a-posteriori (MAP) assignment of latent variables.

3: Yet, standard methods for estimating the MAP assignment do not have a finite time guarantee that the algorithm has converged to a fixed point.

4: Previous research has found that MAP inference can be represented in dual form as a linear programming problem with a non-polynomial number of constraints.

5: A Lagrangian relaxation of the dual yields a statistical inference algorithm as a linear programming problem.

6: However, the decision as to which constraints to remove in the relaxation is often heuristic.

7: We present a method for maximum a-posteriori inference in general Bayesian factor models that sequentially adds constraints to the fully relaxed dual problem using Benders' decomposition.

8: Our method enables the incorporation of expressive integer and logical constraints in clustering problems such as must-link, cannot-link, and a minimum number of whole samples allocated to each cluster.

9: Using this approach, we derive MAP estimation algorithms for the Bayesian Gaussian mixture model and latent Dirichlet allocation.

10: Empirical results show that our method produces a higher optimal posterior value compared to Gibbs sampling and variational Bayes methods for standard data sets and provides certificate of convergence.

11: \end{abstract}

12: