The Annals of Statistics

Differential Geometry of Curved Exponential Families-Curvatures and Information Loss

Shun-Ichi Amari

Full-text: Open access

Abstract

The differential-geometrical framework is given for analyzing statistical problems related to multi-parameter families of distributions. The dualistic structures of the exponential families and curved exponential families are elucidated from the geometrical viewpoint. The duality connected by the Legendre transformation is thus extended to include two kinds of affine connections and two kinds of curvatures. The second-order information loss is calculated for Fisher-efficient estimators, and is decomposed into the sum of two non-negative terms. One is related to the exponential curvature of the statistical model and the other is related to the mixture curvature of the estimator. Only the latter term depends on the estimator, and vanishes for the maximum-likelihood estimator. A set of statistics which recover the second-order information loss are given. The second-order efficiency also is obtained. The differential geometry of the function space of distributions is discussed.

Article information

Source
Ann. Statist. Volume 10, Number 2 (1982), 357-385.

Dates
First available in Project Euclid: 12 April 2007

Permanent link to this document
http://projecteuclid.org/euclid.aos/1176345779

JSTOR
links.jstor.org

Digital Object Identifier
doi:10.1214/aos/1176345779

Mathematical Reviews number (MathSciNet)
MR653513

Zentralblatt MATH identifier
0507.62026

Subjects
Primary: 62E20: Asymptotic distribution theory
Secondary: 62B10: Information-theoretic topics [See also 94A17]

Keywords
Statistical curvatures statistical affine connections information loss recovery of information duality second-order efficiency Kullback-Leibler distance asymptotic estimation theory

Citation

Amari, Shun-Ichi. Differential Geometry of Curved Exponential Families-Curvatures and Information Loss. The Annals of Statistics 10 (1982), no. 2, 357--385. doi:10.1214/aos/1176345779. http://projecteuclid.org/euclid.aos/1176345779.


Export citation