Putting MAP Back on the Map

  • Patrick Pletscher
  • Sebastian Nowozin
  • Pushmeet Kohli
  • Carsten Rother
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6835)


Conditional Random Fields (CRFs) are popular models in computer vision for solving labeling problems such as image denoising. This paper tackles the rarely addressed but important problem of learning the full form of the potential functions of pairwise CRFs. We examine two popular learning techniques, maximum likelihood estimation and maximum margin training. The main focus of the paper is on models such as pairwise CRFs, that are simplistic (misspecified) and do not fit the data well. We empirically demonstrate that for misspecified models maximum-margin training with MAP prediction is superior to maximum likelihood estimation with any other prediction method. Additionally we examine the common belief that MLE is better at producing predictions matching image statistics.


Mean Square Error Minimum Mean Square Error Conditional Random Field Image Denoising Markov Blanket 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Besag, J.: Statistical analysis of non-lattice data. The Statistician (1975)Google Scholar
  2. 2.
    Boykov, Y.: Fast approximate energy minimization via graph cuts. PAMI (2001)Google Scholar
  3. 3.
    Dabov, K., Foi, A., Katkovnik, V., Egiazarian, K.: Image denoising by sparse 3-D transform-domain collaborative filtering. IEEE TIP 16(8), 2080–2095 (2007)MathSciNetGoogle Scholar
  4. 4.
    Felzenszwalb, P.F., Girshick, R.B., McAllester, D., Ramanan, D.: Object detection with discriminatively trained part-based models. PAMI 32(9), 1627–1645 (2010)CrossRefGoogle Scholar
  5. 5.
    Koller, D., Friedman, N.: Probabilistic Graphical Models. MIT Press, Cambridge (2009)zbMATHGoogle Scholar
  6. 6.
    Kolmogorov, V.: Convergent tree-reweighted message passing for energy minimization. PAMI 28(10) (2006)Google Scholar
  7. 7.
    Kumar, S., Hebert, M.: Discriminative Random Fields. IJCV 68(2), 179–201 (2006)CrossRefGoogle Scholar
  8. 8.
    Ladicky, L., Russell, C., Kohli, P., Torr, P.H.S.: Associative hierarchical CRFs for object class image segmentation. In: ICCV 2009 (2009)Google Scholar
  9. 9.
    Mairal, J., Bach, F., Ponce, J., Sapiro, G.: Online Learning for Matrix Factorization and Sparse Coding. JMLR 11, 19–60 (2010)MathSciNetGoogle Scholar
  10. 10.
    Martin, D., Fowlkes, C., Tal, D., Malik, J.: A database of human segmented natural images. In: ICCV 2001 (2001)Google Scholar
  11. 11.
    Nowozin, S., Gehler, P.V., Lampert, C.H.: On parameter learning in CRF-based approaches to object class image segmentation. In: Daniilidis, K., Maragos, P., Paragios, N. (eds.) ECCV 2010. LNCS, vol. 6316, pp. 98–111. Springer, Heidelberg (2010)CrossRefGoogle Scholar
  12. 12.
    Robert, C.P.: The Bayesian Choice. From decision Theoretic Foundations to Computational Implementation. Springer, Heidelberg (2001)zbMATHGoogle Scholar
  13. 13.
    Roth, S., Black, M.J.: Fields of Experts. IJCV (2008)Google Scholar
  14. 14.
    Rother, C., Kolomogorov, V., Blake, A.: GrabCut Interactive Foreground Extraction using Iterated Graph Cuts. TOG 23(3), 309–314 (2004)CrossRefGoogle Scholar
  15. 15.
    Ruderman, D.: The statistics of natural images. Comp. in Neural Systems (1994)Google Scholar
  16. 16.
    Samuel, K.G.G., Tappen, M.F.: Learning Optimized MAP Estimates in Continuously-Valued MRF Models. In: CVPR 2009 (2009)Google Scholar
  17. 17.
    Scharstein, D.: Learning Conditional Random Fields for Stereo. In: CVPR 2007 (2007)Google Scholar
  18. 18.
    Schmidt, U., Gao, Q., Roth, S.: A Generative Perspective on MRFs in Low-Level Vision. In: CVPR 2010 (2010)Google Scholar
  19. 19.
    Taskar, Guestrin, Koller: Max-Margin Markov Networks. In: NIPS 2003 (2003)Google Scholar
  20. 20.
    White, H.: Maximum-likelihood estimation of misspecified models. Econom. (1982)Google Scholar
  21. 21.
    Woodford, O.J., Rother, C., Kolmogorov, V.: A Global Perspective on MAP Inference for Low-Level Vision. In: ICCV 2009 (2009)Google Scholar
  22. 22.
    Zhu, S., Mumford, D.: Prior learning and Gibbs reaction-diffusion. In: PAMI 1997 (1997)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Patrick Pletscher
    • 1
  • Sebastian Nowozin
    • 2
  • Pushmeet Kohli
    • 2
  • Carsten Rother
    • 2
  1. 1.ETH ZurichSwitzerland
  2. 2.Microsoft Research CambridgeUK

Personalised recommendations