Searcharxiv⌕ Search

arXiv subjects

Satyajit Dhadumia

Publications and source records attributed to Satyajit Dhadumia.

3 recordsLinked to original sources

Generalized Cramér-Rao Inequalities Associated with $(h,τ)$-Divergence and $(h,τ)$-Exponential Family

The Kullback-Leibler (KL) divergence and exponential family have laid the foundation for information geometry and have been widely generalized. Using Eguchi's theory, the Fisher information metric can be derived from the KL divergence. Based on this metric, Amari and Nagaoka established an information-geometric proof of the classical Cramér-Rao (CR) inequality for finite-support models. Among the various generalizations of the KL divergence, this paper focuses on the $(h,τ)$-divergence and establishes the associated CR inequality, referred to as the $(h,τ)$-CR inequality, using information geometry. We also provide a proof of this inequality based on the Cauchy-Schwarz inequality, thereby extending its applicability to models with more general support sets. Under suitable regularity conditions, equality holds in the classical CR inequality if and only if the underlying family is exponential. However, for the $(h,τ)$-CR inequality, this property does not generally extend to the $(h,τ)$-exponential family. We therefore establish an attainable CR inequality for this family. This new inequality extends the attainment result of the classical CR inequality by the sufficient statistic of the exponential family to the $(h,τ)$-exponential family.

math.ST↗

Cramer-Rao Type Bounds for Generalized Exponential Families

It is well known that the sufficient statistics of an exponential family attain the usual Cramer-Rao Lower Bound (CRLB) for their expected value. In this paper, we extend this result to a broader family of distributions that includes exponential families and power-law families, with Student t-distributions as special case. To derive the generalized Cramer-Rao inequality, we first apply Eguchi's theory to Bregman divergence to obtain a Riemannian metric and then follow the approach of Amari and Nagaoka. Since this approach applies only to models with finite support, we use Cauchy-Schwarz inequality to extend it to models with countably infinite or continuous support. The resulting generalized CRLB provides a finite-sample lower bound related to the asymptotic covariance matrix of the Minimum Bregman Divergence Estimator (MBDE). We show that the sufficient statistic of the generalized family attains the generalized CRLB. Moreover, although the generalized CRLB does not, in general, coincide with the asymptotic covariance matrix of the MBDE, the covariance matrix of an estimator attaining the generalized CRLB coincides with the corresponding matrix for the joint distribution of the sample.

math.ST↗

A Generalized Cramér-Rao Bound Using Information Geometry

In information geometry, statistical models are considered as differentiable manifolds, where each probability distribution represents a unique point on the manifold. A Riemannian metric can be systematically obtained from a divergence function using Eguchi's theory (1992); the well-known Fisher-Rao metric is obtained from the Kullback-Leibler (KL) divergence. The geometric derivation of the classical Cramér-Rao Lower Bound (CRLB) by Amari and Nagaoka (2000) is based on this metric. In this paper, we study a Riemannian metric obtained by applying Eguchi's theory to the Basu-Harris-Hjort-Jones (BHHJ) divergence (1998) and derive a generalized Cramér-Rao bound using Amari-Nagaoka's approach. There are potential applications for this bound in robust estimation.

math.ST↗