Geometric Characteristics of the Wasserstein Metric on SPD(n) and Its Applications on Data Processing

The Wasserstein distance, especially among symmetric positive-definite matrices, has broad and deep influences on the development of artificial intelligence (AI) and other branches of computer science. In this paper, by involving the Wasserstein metric on SPD(n), we obtain computationally feasible expressions for some geometric quantities, including geodesics, exponential maps, the Riemannian connection, Jacobi fields and curvatures, particularly the scalar curvature. Furthermore, we discuss the behavior of geodesics and prove that the manifold is globally geodesic convex. Finally, we design algorithms for point cloud denoising and edge detecting of a polluted image based on the Wasserstein curvature on SPD(n). The experimental results show the efficiency and robustness of our curvature-based methods.

[1]  Ehsan Nadernejad.  Edge Detection Techniques : Evaluations and Comparisons , 2008 .

[2]  Linyu Peng,et al.  MIG median detectors with manifold filter , 2021, Signal Process..

[3]  Shotaro Akaho,et al.  Learning algorithms utilizing quasi-geodesic flows on the Stiefel manifold , 2005, Neurocomputing.

[4]  A. Ward.  A general analysis of Sylvester's matrix equation , 1991 .

[5]  C. Givens,et al.  A class of Wasserstein metrics for probability distributions. , 1984 .

[6]  Léon Bottou,et al.  Wasserstein GAN , 2017, ArXiv.

[7]  Pierre-Antoine Absil,et al.  Quotient Geometry with Simple Geodesics for the Manifold of Fixed-Rank Positive-Semidefinite Matrices , 2020, SIAM J. Matrix Anal. Appl..

[8]  Asuka Takatsu.  On Wasserstein geometry of Gaussian measures , 2010 .

[9]  Suvrit Sra,et al.  Matrix Manifold Optimization for Gaussian Mixtures , 2015, NIPS.

[10]  E. Masry.  Multivariate probability density estimation for associated processes : Strong consistency and rates , 2002 .

[11]  B. Hall.  Lie Groups, Lie Algebras, and Representations , 2003 .

[12]  Angelika Mueller,et al.  Transformation Groups In Differential Geometry , 2016 .

[13]  John M. Lee.  Introduction to Smooth Manifolds , 2002 .

[14]  John M. Lee.  Riemannian Manifolds: An Introduction to Curvature , 1997 .

[15]  Huafei Sun,et al.  A New Riemannian Structure in SPD(n) , 2019, 2019 IEEE International Conference on Signal, Information and Data Processing (ICSIDP).

[16]  Kon Max Wong,et al.  Riemannian Distances for Signal Classification by Power Spectral Density , 2013, IEEE Journal of Selected Topics in Signal Processing.

[17]  M. Gelbrich.  On a Formula for the L2 Wasserstein Metric between Measures on Euclidean and Hilbert Spaces , 1990 .

[18]  Luigi Malagò,et al.  Wasserstein Riemannian geometry of Gaussian densities , 2018, Information Geometry.

[19]  Jonathan H. Manton,et al.  A Primer on Stochastic Differential Geometry for Signal Processing , 2013, IEEE Journal of Selected Topics in Signal Processing.

[20]  Asuka Takatsu.  Wasserstein geometry of Gaussian measures , 2011 .

[21]  Leonidas J. Guibas,et al.  A metric for distributions with applications to image databases , 1998, Sixth International Conference on Computer Vision (IEEE Cat. No.98CH36271).

[22]  Hongqiang Wang,et al.  Target Detection Within Nonhomogeneous Clutter Via Total Bregman Divergence-Based Matrix Information Geometry Detectors , 2020, IEEE Transactions on Signal Processing.

[23]  Xavier Pennec,et al.  A Riemannian Framework for Tensor Computing , 2005, International Journal of Computer Vision.

[24]  Radu Bogdan Rusu,et al.  3D is here: Point Cloud Library (PCL) , 2011, 2011 IEEE International Conference on Robotics and Automation.

[25]  N. Ayache,et al.  Log‐Euclidean metrics for fast and simple calculus on diffusion tensors , 2006, Magnetic resonance in medicine.

[26]  W. Boothby.  An introduction to differentiable manifolds and Riemannian geometry , 1975 .

[27]  Simone G. O. Fiori,et al.  A Theory for Learning by Weight Flow on Stiefel-Grassman Manifold , 2001, Neural Computation.

[28]  R. Bhatia.  Positive Definite Matrices , 2007 .