Generalized sequential tree-reweighted message passing

preprint Generalized sequential tree-reweighted message passing submitted Vladimir Kolmogorov author 3D50B0BA-F248-11E8-B48F-1D18A9856A87 Thomas Schoenemann author VlKo department This paper addresses the problem of approximate MAP-MRF inference in general graphical models. Following [36], we consider a family of linear programming relaxations of the problem where each relaxation is specified by a set of nested pairs of factors for which the marginalization constraint needs to be enforced. We develop a generalization of the TRW-S algorithm [9] for this problem, where we use a decomposition into junction chains, monotonic w.r.t. some ordering on the nodes. This generalizes the monotonic chains in [9] in a natural way. We also show how to deal with nested factors in an efficient way. Experiments show an improvement over min-sum diffusion, MPLP and subgradient ascent algorithms on a number of computer vision and natural language processing problems. 2012 eng arXiv 1205.635210.48550/arXiv.1205.6352 Kolmogorov, Vladimir, and Thomas Schoenemann. “Generalized Sequential Tree-Reweighted Message Passing.” ArXiv, 1205.6352, doi:<a href="https://doi.org/10.48550/arXiv.1205.6352">10.48550/arXiv.1205.6352</a>. Kolmogorov V, Schoenemann T. Generalized sequential tree-reweighted message passing. arXiv. doi:<a href="https://doi.org/10.48550/arXiv.1205.6352">10.48550/arXiv.1205.6352</a> Kolmogorov, Vladimir, and Thomas Schoenemann. “Generalized Sequential Tree-Reweighted Message Passing.” ArXiv, n.d. <a href="https://doi.org/10.48550/arXiv.1205.6352">https://doi.org/10.48550/arXiv.1205.6352</a>. Kolmogorov V, Schoenemann T. Generalized sequential tree-reweighted message passing. arXiv, 1205.6352. V. Kolmogorov and T. Schoenemann, “Generalized sequential tree-reweighted message passing,” arXiv. . Kolmogorov, V., & Schoenemann, T. (n.d.). Generalized sequential tree-reweighted message passing. arXiv. <a href="https://doi.org/10.48550/arXiv.1205.6352">https://doi.org/10.48550/arXiv.1205.6352</a> V. Kolmogorov, T. Schoenemann, ArXiv (n.d.). 29282018-12-11T12:00:23Z2025-07-08T07:19:42Z