The Riemannian geometry of certain parameter estimation problems with singular Fisher information matrices

Abstract

Many parametric statistical models suffer from "intrinsic ambiguities" in the sense that the distribution of the observation vector is invariant to smooth, structured changes in the model's parameters. The fact that certain members of the parametric statistical family are locally undistinguishable makes the Fisher information matrix (FIM) associated to the given statistical model singular. We examine such degenerate deterministic parameter estimation problems from a Riemannian geometric perspective. We start by replacing the original (ambiguous) parameter set by a lower-dimensional Riemannian (non-ambiguous) parameter set. The new parameter set comes in the form of a quotient space and is obtained by identifying equivalent family members in the initial parameterization. We specialize recently developed extensions of the Cramér-Rao bound (CRB) for the Riemannian setup to this particular setting. This offers a re-interpretation of the CRB inequality involving the pseudo-inverse of the FIM. Also, we present a lower bound on the variance (computed with respect to the geodesic distance) of unbiased estimators taking values in the quotient space. Geometrically, this corresponds to a fundamental limit on the capability of these estimators in discriminating adjacent parameter equivalence classes in the original problem parameterization. A numerical example involving the blind identification of single-input multipleoutput (SIMO) channels driven by a Gaussian source is worked out.

Intrinsic Variance Lower Bound (IVLB): An Extension of the Cramer-Rao Bound to Riemannian Manifolds

“…This viewpoint has also advantages in practical identification algorithms as first noticed by Manton [11]. In the context of performance analysis, see [10] for other applications of such constructions.…”

A Novel Method for Reconstructing 3D Scattering Centers Based on Multiple HRR Profiles and Its Performance Bounds

“…Recent results in information theory [6][7][8][9] are used in deducing its CRB. We present assumptions and main conclusions here, and more details are referred to the literatures.…”

“…To further evaluate the performance of SSI method, we derive the Cramer-Rao Lower Bounds (CRB) of the reconstruction problem. Since it is essentially an estimation problem with constraints and unidentifiable parameters, we resort to recent results in CRB theory about singular Fisher information matrix (FIM) [6][7][8][9]. Simulations show that SSI algorithm approaches CRB at ordinary SNRs.…”

A Fresh Look at the Bayesian Bounds of the Weiss-Weinstein Family

“…In this paper, we will first show that an appropriate choice of leads to the Bayesian bounds of the Weiss-Weinstein family. Second, we will show how this approach can be used in order to build new minimal (10) and and (11) and and and bounds, particularly, by solving the following constrained optimization problem: (16) In this section, we restrict , and in order to obtain a general framework to create minimal bounds. Then, by way of a constrained optimization problem for which we give an explicit solution we unify the bounds of the Weiss-Weinstein family.…”