statistics on lie groups
play

Statistics on Lie groups: using the pseudo-Riemannian framework? - PowerPoint PPT Presentation

Statistics on Lie groups: using the pseudo-Riemannian framework? Nina Miolane, Xavier Pennec Asclepios Team, INRIA Sophia Antipolis Max Ent 2014 Nina Miolane - Statistics on Lie groups Models with Lie groups Articulated models Shape


  1. Statistics on Lie groups: using the pseudo-Riemannian framework? Nina Miolane, Xavier Pennec Asclepios Team, INRIA Sophia Antipolis Max Ent ’ 2014 Nina Miolane - Statistics on Lie groups

  2. Models with Lie groups Articulated models Shape models Robotics Computational Medicine Reference  1 𝑻𝑷 πŸ’ Spherical arm  5 Patient 1  2  4 Patient 5  3 http://www.societyofrobots.com/ Patient 4 Patient 3 Patient 2 Paleontology Computational Anatomy 𝑻𝑭(πŸ’) 2006, Nature Publishing Group, Genetics οƒ  Statistics 10/17/2014 2 Nina Miolane - Statistics on Lie groups

  3. Models with Lie groups Articulated model of spine: Statistics on spines: Vertebra 𝑆, 𝑒 ∈ 𝑇𝐹(3) Mean, PCA… Rotation Matrix translation vector (𝑆, 𝑒) PCA : Modes 1 to 4 Computational Anatomy : model and analyze the variability of human anatomy 10/17/2014 3 Nina Miolane - Statistics on Lie groups

  4. New statistics on Lie groups? On Lie group 𝑻𝑭(πŸ’) Consequence for model Vertebra : 𝑆, 𝑒 ∈ 𝑇𝐹(3) (space of transformations) 𝐒 𝟐 , 𝒖 𝟐 𝐒 πŸ‘ , 𝒖 πŸ‘ ? ( 𝑺 𝟐 +𝑺 πŸ‘ , 𝒖 𝟐 +𝒖 πŸ‘ 𝐒 𝟐 , 𝒖 𝟐 𝑺 𝟐 +𝑺 πŸ‘ 𝒖 𝟐 +𝒖 πŸ‘ πŸ‘ ) ( πŸ‘ ) , πŸ‘ πŸ‘ ? 𝐒 πŸ‘ , 𝒖 πŸ‘ 𝑻𝑭 πŸ’ Mean not on 𝑇𝐹(3) , thus not a Average pose of two vertebrae is rigid transformation not a vertebra Usual statistics: linear Need new statistics! Lie groups: not linear in general 10/17/2014 4 Nina Miolane - Statistics on Lie groups

  5. Outline Can we use the pseudo-Riemannian framework to define statistics on Lie groups? 1. Pseudo-Riemannian structures on Lie groups 2. Algorithm to compute bi-invariant pseudo-metrics 3. Results on selected Lie groups 10/17/2014 5 Nina Miolane - Statistics on Lie groups

  6. Outline Can we use the pseudo-Riemannian framework to define statistics on Lie groups? 1. Pseudo-Riemannian structures on Lie groups 2. An algorithm to compute bi-invariant pseudo-metrics 3. Results on selected Lie groups 10/17/2014 6 Nina Miolane - Statistics on Lie groups

  7. (Pseudo-) Riemannian structure on Lie groups Manifold 𝑡 Differential structure + algebraic + metrical Lie group (Pseudo-) Riemannian manifold (𝑯,∘) (𝑡, <, >) Metric <, > : collection of positive definite inner products Pseudo-metric <, > : collection of definite inner products Consistency of the structures requires bi-invariant pseudo-metric Quadratic Lie group (𝑯,∘, <, >) 10/17/2014 7 Nina Miolane - Statistics on Lie groups

  8. Statistics on Lie groups : the mean 𝒉 𝑕 𝑗 𝑗 οƒ  Requirements for mean of data set Consistency Computability Riemannian structure 𝒉 mean of 𝑕 𝑗 𝑗 οƒ  Bi-invariant metric Conditions on 𝑕 𝑗 𝑗 FrΓ©chet mean = group mean 𝑀 β„Ž such that exists 𝒉 FrΓ©chet mean: definition with and is computability conditions unique? π’Š ∘ 𝒉 mean of β„Ž ∘ 𝑕 𝑗 𝑗 Group exponential barycenter [1] =group mean [1] Pennec & Arsigny. Exponential barycenter of the canonical Cartan connection and invariant means on Lie groups. (2012) 10/17/2014 8 Nina Miolane - Statistics on Lie groups

  9. Bi-invariant metric? On all Lie groups? NO! Existence of a bi-invariant metric ? On Lie groups that have a group mean? NO! Characterization of Lie groups with 𝑻𝑭(𝒐) : bi-invariant metric by Cartan [2] : β€’ unique (a.e.) group mean compact & abelian β€’ no bi-invariant metric οƒ  Group mean not characterized by a bi-invariant metric [1] [1] Pennec & Arsigny. Exponential barycenter of the canonical Cartan connection and invariant means on Lie groups. (2012) [2] Elie Cartan. La thΓ©orie des groups finis et continus et l’analyse situs. (1952) 10/17/2014 9 Nina Miolane - Statistics on Lie groups

  10. Statistics on Lie groups : the mean 𝒉 𝑕 𝑗 𝑗 οƒ  Requirements for mean of data set Consistency Computability Riemannian structure 𝒉 mean of 𝑕 𝑗 𝑗 FrΓ©chet mean: definition with Conditions computability conditions on 𝑕 𝑗 𝑗 𝑀 β„Ž such that exists 𝒉 Bi-invariant metric and is οƒ  FrΓ©chet mean = group mean unique? π’Š ∘ 𝒉 mean of β„Ž ∘ 𝑕 𝑗 𝑗 Generalize Riemannian to pseudo-Riemannian Pseudo-Riemannian structure Group exponential barycenter [1] =group mean Bi-invariant pseudo-metric [1] Pennec & Arsigny. Exponential barycenter of the canonical Cartan connection and invariant means on Lie groups. (2012) 10/17/2014 10 Nina Miolane - Statistics on Lie groups

  11. Bi-invariant pseudo-metrics? On all Lie groups? NO! Existence of a bi-invariant pseudo-metric ? On Lie groups that have a group mean? 𝑇𝐹(π‘œ) : Characterization of Lie groups with β€’ unique (a.e.) group mean bi-invariant pseudo-metric β€’ bi-invariant pseudo-metric ? by Medina & Revoy [3] Group mean characterized by a bi-invariant pseudo-metric ? [3] Medina & Revoy. Groupes de Lie munis de pseudo-metriques bi-invariantes. (1982) 10/17/2014 11 Nina Miolane - Statistics on Lie groups

  12. Outline Can we use the pseudo-Riemannian framework to define statistics on Lie groups? 1. Pseudo-Riemannian structures on Lie groups 2. An algorithm to compute bi-invariant pseudo-metrics 3. Results on selected Lie groups 10/17/2014 12 Nina Miolane - Statistics on Lie groups

  13. From Lie group to Lie algebra Compute bi-invariant pseudo-metrics on finite dimensional connected Lie group 𝑯 𝑬𝑴 π’Š . π’˜ bi-invariant : < 𝒗, π’˜ > 𝒉 =< 𝑬𝑴 π’Š . 𝒗, 𝑬𝑴 π’Š . π’˜ > 𝑴 π’Š 𝒉 =< 𝐸𝑆 β„Ž . 𝑣, 𝐸𝑆 β„Ž . 𝑀 > 𝑆 β„Ž 𝑕 𝑴 π’Š where 𝑀 β„Ž 𝑕 = β„Ž ∘ 𝑕 and 𝑆 β„Ž 𝑕 = 𝑕 ∘ β„Ž 𝒗 π’˜ 𝑴 π’Š 𝒉 = π’Š ∘ 𝒉 𝒉 𝑬𝑴 π’Š . 𝒗 Compute bi-invariant pseudo-metrics on its Lie algebra 𝖍 = 𝐔 𝐟 𝐇, +, . , , 𝖍 bi-invariant: < 𝑦, 𝑧 π”₯ , 𝑨 > 𝑓 +< 𝑧, 𝑦, 𝑨 π”₯ > 𝑓 = 0 Structure of quadratic 𝖍 ? 10/17/2014 13 Nina Miolane - Statistics on Lie groups

  14. Lie algebra representations Structure of quadratic 𝖍 ? Study adjoint representation of π”₯ Representation 𝜽 of 𝖍 on 𝑾 : Lie algebra homomorphism πœƒ: π”₯ ↦ π”₯π”ͺ(π‘Š) Ex.: Homogeneous representation of 𝑇𝐹 3 on ℝ 4 : πœƒ: 𝑇𝐹 3 ↦ π”₯π”ͺ ℝ 4 1 . 𝑦 s.t. 𝑆 𝑒 1 = 𝑆. 𝑦 + 𝑒 𝑆, 𝑒 ↦ 𝑆 𝑒 0 1 1 0 Subrepresentation: subspace of π‘Š stable by πœƒ π”₯ Subrepresentation decomposition: π‘Š = 𝐢 1 βŠ• π”₯ … βŠ• π”₯ 𝐢 𝑂 with 𝐢 𝑗 subrepresentations Indecomposable subrepresentation: not a sum of subrepresentations Adjoint representation 𝒃𝒆 of 𝖍 (on itself: 𝑾 = 𝖍 ) 𝑏𝑒: π”₯ ↦ π”₯π”ͺ(π”₯) 𝑦 ↦ 𝑏𝑒 𝑦 = 𝑦,βˆ™ π”₯ π”₯ βŠ• π”₯ … βŠ• π”₯ 𝐢 𝑂 π”₯ : decomposition into indecomposable subrepresentations π”₯ = 𝐢 1 10/17/2014 14 Nina Miolane - Statistics on Lie groups

  15. Structure of quadratic π”₯ π”₯ π”₯ π”₯ π”₯ 𝐢 1 𝐢 𝑗 𝐢 𝑂 𝖍 = 𝐓 βŠ• 𝑻 βˆ— 𝖍 = 𝐗 βŠ• 𝐓 βŠ• 𝑻 βˆ— 𝖍 =1-dim. 𝖍 =simple π‘ͺ 𝒋 π‘ͺ 𝒋 π‘ͺ 𝒋 π‘ͺ 𝒋 Th. (Medina & Revoy): Structure of quadratic 𝖍 π”₯ βŠ• π”₯ … βŠ• π”₯ B N π”₯ has indecomposable B 𝑗 π”₯ s.t.: Adjoint representation decomposition π”₯ = B 1 𝖍 simple or 1-dim. Type (1): π‘ͺ 𝒋 𝖍 = 𝐗 βŠ• 𝐓 βŠ• 𝐓 βˆ— double extension of 𝑋 quadratic by 𝑇 of Type (1) Type (2): π‘ͺ 𝒋 10/17/2014 15 Nina Miolane - Statistics on Lie groups

  16. Bi-invariant pseudo-metric on quadratic π”₯ β€² + β‹― + 𝑐 𝑂 > π”₯ π”₯ < 𝑐 1 + β‹― + 𝑐 𝑂 , 𝑐 1 β€² > 𝐢 1 + β‹― +< 𝑐 𝑂 , 𝑐 𝑂 β€² > 𝐢 𝑂 =< 𝑐 1 , 𝑐 1 π”₯ π”₯ π”₯ 𝐢 1 𝐢 𝑗 𝐢 𝑂 𝖍 = 𝐓 βŠ• 𝑻 βˆ— 𝖍 = 𝐗 βŠ• 𝐓 βŠ• 𝑻 βˆ— 𝖍 =1-dim. 𝖍 =simple π‘ͺ 𝒋 π‘ͺ 𝒋 π‘ͺ 𝒋 π‘ͺ 𝒋 𝖍 = π‘³π’‹π’Žπ’Žπ’‹π’π’‰(𝒄, 𝒄 β€² ) < 𝒄, 𝒄′ > π‘ͺ 𝒋 < 𝒙 + 𝒕 + π’ˆ, 𝒙 β€² + 𝒕 β€² + π’ˆ β€² > π‘ͺ 𝒋 𝖍 = < 𝒙, 𝒙 β€² > 𝑿 + π’ˆ 𝒕 β€² + π’ˆβ€²(𝒕) < 𝒄, 𝒄 β€² > π‘ͺ 𝒋 𝖍 = 𝒄. 𝒄′ < 𝒕 + π’ˆ, 𝒕 β€² + π’ˆ β€² > π‘ͺ 𝒋 𝖍 = π’ˆ 𝒕 β€² + π’ˆβ€²(𝒕) 10/17/2014 16 Nina Miolane - Statistics on Lie groups

  17. Algorithm π”₯ π”₯ π”₯ π”₯ 𝐢 1 𝐢 𝑂 𝐢 𝑗 𝖍 = 𝐓 βŠ• 𝑻 βˆ— ? 𝖍 = 𝐗 βŠ• 𝐓 βŠ• 𝑻 βˆ— ? Else : EXIT 𝖍 =1-dim. ? 𝖍 =simple? π‘ͺ 𝒋 π‘ͺ 𝒋 π‘ͺ 𝒋 π‘ͺ 𝒋 If algorithm finishes: expression of a bi-invariant pseudo-metric on π”₯ If EXIT: no bi-invariant pseudo-metric on π”₯ 10/17/2014 17 Nina Miolane - Statistics on Lie groups

  18. Outline Can we use the pseudo-Riemannian framework to define statistics on Lie groups? 1. Pseudo-Riemannian structures on Lie groups 2. An algorithm to compute bi-invariant pseudo-metrics 3. Results on selected Lie groups 10/17/2014 18 Nina Miolane - Statistics on Lie groups

Recommend


More recommend