A Novel Tree Block-Coordinate Method for MAP Inference

  • Christopher Zach
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 9358)


Block-coordinate methods inspired by belief propagation are among the most successful methods for approximate MAP inference in graphical models. The set of unknowns optimally updated in such block-coordinate methods is typically very small and spans only single edges or shallow trees. We derive a method that optimally updates sets of unknowns spanned by an arbitrary tree that is different from one reported in the literature. It provides some insight why “tree block-coordinate” methods are not as useful as expected, and enables a simple technique to makes these tree updates more effective.

Supplementary material (678 kb)
Supplementary material 1 (zip 678 KB)


  1. 1.
    Boykov, Y., Veksler, O., Zabih, R.: Fast approximate energy minimization via graph cuts. IEEE Trans. Pattern Anal. Mach. Intell. 23(11), 1222–1239 (2001)CrossRefGoogle Scholar
  2. 2.
    Domke, J.: Parameter learning with truncated message-passing. In: Proceedings of the CVPR, pp. 2937–2943 (2011)Google Scholar
  3. 3.
    Drory, A., Haubold, C., Avidan, S., Hamprecht, F.A.: Semi-global matching: a principled derivation in terms of message passing. In: Jiang, X., Hornegger, J., Koch, R. (eds.) GCPR 2014. LNCS, vol. 8753, pp. 43–53. Springer, Heidelberg (2014) Google Scholar
  4. 4.
    Globerson, A., Jaakkola, T.: Fixing max-product: convergent message passing algorithms for MAP LP-relaxations. In: NIPS (2007)Google Scholar
  5. 5.
    Hazan, T., Shashua, A.: Norm-prodcut belief propagtion: primal-dual message-passing for LP-relaxation and approximate-inference. IEEE Trans. Inf. Theory 56(12), 6294–6316 (2010)CrossRefzbMATHGoogle Scholar
  6. 6.
    Hirschmüller, H.: Accurate and efficient stereo processing by semi-global matching and mutual information. In: Proceedings CVPR, pp. 807–814 (2005)Google Scholar
  7. 7.
    Kolmogorov, V.: Convergent tree-reweighted message passing for energy minimization. IEEE Trans. Pattern Anal. Mach. Intell. 28(10), 1568–1583 (2006)CrossRefGoogle Scholar
  8. 8.
    Lempitsky, V., Rother, C., Roth, S., Blake, A.: Fusion moves for Markov random field optimization. IEEE Trans. Pattern Anal. Mach. Intell. 32(8), 1392–1405 (2010)CrossRefGoogle Scholar
  9. 9.
    Pearl, J.: Probabilistic Reasoning in Intelligent Systems. Morgan Kaufmann, Mountain view (1988)zbMATHGoogle Scholar
  10. 10.
    Scharstein, D., Szeliski, R.: High-accuracy stereo depth maps using structured light. In: Proceedings of the CVPR, pp. 195–202 (2003)Google Scholar
  11. 11.
    Sontag, D., Globerson, A., Jaakkola, T.: Optimization for machine learning. In: Introduction to Dual Decomposition for Inference. MIT Press (2011)Google Scholar
  12. 12.
    Sontag, D., Jaakkola, T.: Tree block coordinate descent for MAP in graphical models. J. Mach. Learn. Res. 5, 544–551 (2009)Google Scholar
  13. 13.
    Wainwright, M.J., Jordan, M.I.: Graphical models, exponential families, and variational inference. Found. Trends Mach. Learn. 1, 1–305 (2008)CrossRefzbMATHGoogle Scholar
  14. 14.
    Wainwright, M.J., Jaakkola, T.S., Willsky, A.S.: MAP estimation via agreement on trees: message-passing and linear programming. IEEE Trans. Inf. Theory 51(11), 3697–3717 (2005)MathSciNetCrossRefzbMATHGoogle Scholar
  15. 15.
    Weiss, Y., Yanover, C., Meltzer, T.: MAP estimation, linear programming and belief propagation with convex free energies. In: UAI (2007)Google Scholar
  16. 16.
    Werner, T.: A linear programming approach to max-sum problem: a review. IEEE Trans. Pattern Anal. Mach. Intell. 29(7), 1165–1179 (2007)CrossRefGoogle Scholar
  17. 17.
    Zach, C.: A principled approach for coarse-to-fine MAP inference. In: Proceedings of the CVPR, pp. 1330–1337 (2014)Google Scholar

Copyright information

© Springer International Publishing Switzerland 2015

Open Access This chapter is distributed under the terms of the Creative Commons Attribution Noncommercial License, which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.

Authors and Affiliations

  1. 1.Toshiba Research EuropeCambridgeUK

Personalised recommendations