minnorm approximation of maxent mindiv problems for
play

MinNorm approximation of MaxEnt/MinDiv problems for probability - PowerPoint PPT Presentation

MinNorm approximation of MaxEnt/MinDiv problems for probability tables Patrick Bogaert and Sarah Gengler UCL Rebuilding probability tables UCL Rebuilding probability tables Limited number of samples Poor estimates UCL Rebuilding


  1. MinNorm approximation of MaxEnt/MinDiv problems for probability tables Patrick Bogaert and Sarah Gengler UCL

  2. Rebuilding probability tables UCL

  3. Rebuilding probability tables • Limited number of samples  Poor estimates UCL

  4. Rebuilding probability tables • Limited number of samples  Poor estimates • How to integrate experts opinion ? UCL

  5. Rebuilding probability tables • Limited number of samples  Poor estimates • How to integrate experts opinion ? Rewriting information as equality / inequality constraints  UCL

  6. Rebuilding probability tables • Limited number of samples  Poor estimates • How to integrate experts opinion ? Rewriting information as equality / inequality constraints  UCL

  7. Rebuilding probability tables • Limited number of samples  Poor estimates • How to integrate experts opinion ? Rewriting information as equality / inequality constraints  • Equality constraints MaxEnt  • Inequality constraints Minimum divergence ( MinDiv )  UCL

  8. Rebuilding probability tables • Limited number of samples  Poor estimates • How to integrate experts opinion ? Rewriting information as equality / inequality constraints  • Equality constraints MaxEnt  • Inequality constraints Minimum divergence ( MinDiv )   Need for an efficient methodology to rebuild probability tables from both equality and inequality constraints UCL

  9. The MaxEnt problem UCL

  10. The MaxEnt problem • Equality constraints UCL

  11. The MaxEnt problem • Equality constraints • Entropy maximized subject to the equality constraints UCL

  12. The MaxEnt problem • Equality constraints • Entropy maximized subject to the equality constraints UCL

  13. The MaxEnt problem • Equality constraints • Entropy maximized subject to the equality constraints  Sequence of MinNorm problems for solving the MaxEnt problem UCL

  14. MinNorm as an approximation of MaxEnt UCL

  15. MinNorm as an approximation of MaxEnt • Taylor series of ln p i around p i = k i UCL

  16. MinNorm as an approximation of MaxEnt • Taylor series of ln p i around p i = k i • Truncating at degree one and summing over i UCL

  17. MinNorm as an approximation of MaxEnt • Taylor series of ln p i around p i = k i • Truncating at degree one and summing over i • In particular, if k i = 1/n UCL

  18. MinNorm as an approximation of MaxEnt • For any other choice of the k i ‘s, by completing the square UCL

  19. MinNorm as an approximation of MaxEnt • For any other choice of the k i ‘s, by completing the square • Summing over i Where UCL

  20. The MinDiv problem UCL

  21. The MinDiv problem • Divergence or Kullback-Leibler distance UCL

  22. The MinDiv problem • Divergence or Kullback-Leibler distance • Equality constraints = 0 Maximizing UCL

  23. The MinDiv problem • Divergence or Kullback-Leibler distance • Equality constraints = 0 Maximizing  UCL

  24. The MinDiv problem • Divergence or Kullback-Leibler distance • Equality constraints = 0 Maximizing   UCL

  25. The MinDiv problem • Divergence or Kullback-Leibler distance • Equality constraints = 0 Maximizing    Sequence of MinNorm problems for solving the MinDiv problem UCL

  26. The MinDiv problem • Divergence or Kullback-Leibler distance • Equality constraints = 0 Maximizing    Sequence of MinNorm problems for solving the MinDiv problem Both Equality and Inequality constraints can be processed together by MinNorm approximations UCL

  27. MinNorm as an approximation of MinDiv UCL

  28. MinNorm as an approximation of MinDiv • Taylor series around p i = k i and completing the square UCL

  29. MinNorm as an approximation of MinDiv • Taylor series around p i = k i and completing the square • Summing over i Where UCL

  30. Application in drainage classes mapping UCL

  31. Application in drainage classes mapping • Categorical data are found in a wide variety of applications UCL

  32. Application in drainage classes mapping UCL

  33. Application in drainage classes mapping • Categorical data are found in a wide variety of applications • 90 % of variables collected in soil surveys are categorical • Soil drainage, an important criterion in rating soils for various uses UCL

  34. Application in drainage classes mapping UCL

  35. Application in drainage classes mapping UCL

  36. Application in drainage classes mapping UCL

  37. Application in drainage classes mapping UCL

  38. Application in drainage classes mapping UCL

  39. Application in drainage classes mapping UCL

  40. Application in drainage classes mapping UCL

  41. Application in drainage classes mapping UCL

  42. Integrating the lithological map : 4 cases UCL

  43. Integrating the lithological map : 4 cases UCL

  44. Integrating the lithological map : 4 cases UCL

  45. Spatial prediction UCL

  46. Integrating the lithological map : 4 cases UCL

  47. Spatial prediction UCL

  48. Integrating the lithological map : 4 cases UCL

  49. Conclusions UCL

  50. Conclusions UCL

  51. Conclusions  MinNorm Approximations UCL

  52. Conclusions  MinNorm Approximations UCL

  53. Conclusions  MinNorm Approximations UCL

  54. Conclusions  MinNorm Approximations UCL

  55. Conclusions  MinNorm Approximations UCL

  56. Conclusions  MinNorm Approximations UCL

  57. Conclusions  MinNorm Approximations UCL

  58. Thank you for your attention UCL

  59. References UCL

Recommend


More recommend