1 / 113
Télécharger la présentation

An Interesting Question

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. An Interesting Question How generally applicable is Backwards approach to PCA? An Attractive Answer: James Damon, UNC Mathematics Key Idea: Express Backwards PCA as Nested Series of Constraints

  2. General View of Backwards PCA Define Nested Spaces via Constraints E.g. SVD Now Define: ConstraintGives Nested Reduction of Dim’n

  3. Vectors of Angles Vectors of Angles as Data Objects Slice space with hyperplanes???? (ala Principal Nested Spheres)

  4. Vectors of Angles E.g. , Data w/ “Single Mode of Var’n” Best Fitting Planar Slice gives Bimodal Dist’n Special Thanks to Eduardo García-Portugués

  5. Torus Space Try To Fit A Geodesic Challenge: Can Get Arbitrarily Close

  6. Torus Space Fit Nested Sub-Manifold

  7. PNS Main Idea Data Objects: Where is a dimensional manifold Consider a nested series of sub-manifolds: where for and Goal: Fit all of simultaneously to

  8. General Background Call each a stratum, so is a manifold stratification To be fit to New Approach: Simultaneously fit Nested Submanifold (NS)

  9. Projection Notation For let denote the telescoping projection onto I.e. for Note: This projection is fundamental to Backwards PCA methods

  10. PNS Components For a given , represent a point by its Nested Submanifold components: where for In the sense that “” means the shortest geodesic arc between &

  11. Nested Submanifold Fits Simultaneous Fit Criteria? Based on Stratum-Wise Sums of Squares For define Uses “lengths” of NS Components:

  12. NS Components in NS Candidate 2 (Shifted to Sample Mean) Note: Both & Decrease

  13. NS Components in NS based On PC1 Note: Yet is Constant (Pythagorean Thm)

  14. NS Components in NS based On PC2 Note: is Constant (Pythagorean Thm)

  15. NS Components in NS Candidate 1

  16. NS Components in NS Candidate 2

  17. NS Components in NS based On PC1

  18. NS Components in NS based On PC2

  19. Nested Submanifold Fits Simultaneously fit Simultaneous Fit Criterion? Above Suggests Want: Works for Euclidean PCA (?)

  20. Nested Submanifold Fits Simultaneous Fit Criterion? Above Suggests Want: Important Predecessor Pennec(2016) AUC Criterion:

  21. Pennec’s Area Under the Curve Based on Scree Plot 2 Component Index

  22. Pennec’s Area Under the Curve Based on Scree Plot Cumulative 2 Component Index

  23. Pennec’s Area Under the Curve Based on Scree Plot Cumulative Area = Component Index

  24. Torus Space Fit Nested Sub-Manifold Choice of & in: ???

  25. Torus Space Tiled embedding is complicated (maybe OK for low rank approx.) Instead Consider Nested Sub-Torii Work in Progress with Garcia, Wood, Le Key Factor: Important Modes of Variation

  26. OODA Big Picture New Topic: Curve Registration Main Reference: Srivastava et al (2011)

  27. Collaborators • AnujSrivastava(Florida State U.) • Wei Wu (Florida State U.) • Derek Tucker (Florida State U.) • Xiaosun Lu (U. N. C.) • Inge Koch (U. Adelaide) • Peter Hoffmann (U. Adelaide) • J. O. Ramsay (McGill U.) • Laura Sangalli (Milano Polytech.)

  28. Context Functional Data Analysis Curves as Data Objects Toy Example:

  29. Context Functional Data Analysis Curves as Data Objects Toy Example: How Can We Understand Variation?

  30. Context Functional Data Analysis Curves as Data Objects Toy Example: How Can We Understand Variation?

  31. Context Functional Data Analysis Curves as Data Objects Toy Example: How Can We Understand Variation?

  32. Functional Data Analysis Insightful Decomposition

  33. Functional Data Analysis Insightful Decomposition • Horiz’l • Var’n

  34. Functional Data Analysis Insightful Decomposition Vertical Variation • Horiz’l • Var’n

  35. Challenge • Fairly Large Literature • Many (Diverse) Past Attempts • Limited Success (in General) • Surprisingly Slippery (even mathematical formulation)

  36. Challenge (Illustrated) Thanks to Wei Wu

  37. Challenge (Illustrated) Thanks to Wei Wu

  38. Functional Data Analysis Appropriate Mathematical Framework? Vertical Variation • Horiz’l • Var’n

  39. Landmark Based Shape Analysis Approach: Identify objects that are: • Translations • Rotations • Scalings of each other Mathematics: Equivalence Relation Results in: Equivalence Classes Which become the Data Objects

  40. Landmark Based Shape Analysis Equivalence Classes become Data Objects a.k.a. “Orbits” Mathematics: Called “Quotient Space” , , , , , ,

  41. Curve Registration What are the Data Objects? Vertical Variation • Horiz’l • Var’n

  42. Curve Registration What are the Data Objects? Consider “Time Warpings” (smooth) More Precisely: Diffeomorphisms

  43. Curve Registration Diffeomorphisms • is 1 to 1 • is onto (thus is invertible) • Differentiable • is Differentiable

  44. Time Warping Intuition Elastically Stretch & Compress Axis

  45. Time Warping Intuition Elastically Stretch & Compress Axis (identity)

  46. Time Warping Intuition Elastically Stretch & Compress Axis

  47. Time Warping Intuition Elastically Stretch & Compress Axis

  48. Time Warping Intuition Elastically Stretch & Compress Axis

  49. Curve Registration Say curves and are equivalent, When so that

  50. Curve Registration Toy Example: Starting Curve,

More Related