Improving Cramér-Rao Bound With Multivariate Parameters: An Extrinsic Geometry Perspective
By: Sunder Ram Krishnan
Potential Business Impact:
Improves how well we can measure things precisely.
We derive a vector generalization of the square root embedding-based curvature-corrected Cram\'er--Rao bound (CRB) previously considered by the same author in \cite{srk} with scalar parameters. A \emph{directional} curvature correction is established first, and sufficient conditions for a conservative matrix-level CRB refinement are formulated using a simple semidefinite program. The directional correction theorem is rigorously illustrated with a Gaussian example.
Similar Papers
Improving Cramér-Rao Bound And Its Variants: An Extrinsic Geometry Perspective
Statistics Theory
Makes measurements more accurate by understanding shape.
Cartan meets Cramér-Rao
Statistics Theory
Improves how well computers guess things by using geometry.
Cramer-Rao Bounds for Laplacian Matrix Estimation
Machine Learning (Stat)
Helps power grids and computer networks work better.