Properties of discrete Fisher information: Cramer-Rao-type and log-Sobolev-type inequalities
The Fisher information have connections with the standard deviation and the Shannon differential entropy through the Cramer-Rao bound and the log-Sobolev inequality. These inequalities hold for continuous distributions. In this paper, we introduce the Fisher information for discrete distributions (DFI) and show that the DFI satisfies the Cramer-Rao-type bound and the log-Sobolev-type inequality.
READ FULL TEXT