Published In Distributed Parallel Inference on Large Factor Graphs

As computer clusters become more common and the size of the problems encountered in the field of AI grows, there is an increasing demand for efficient parallel inference algorithms. We con-sider the problem of parallel inference on large factor graphs in the distributed memory setting of computer clusters. We develop a new effi-cient parallel inference algorithm, DBRSplash, which incorporates over-segmented graph parti-tioning, belief residual scheduling, and uniform work Splash operations. We empirically evalu-ate the DBRSplash algorithm on a 120 proces-sor cluster and demonstrate linear to super-linear performance gains on large factor graph models.