Differential Geometry of Curved Exponential Families-Curvatures and Information Loss

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.

[1]  Luther Pfahler Eisenhart,et al.  Non-Riemannian Geometry , 2005 .

[2]  H. Chernoff.  A Measure of Asymptotic Efficiency for Tests of a Hypothesis Based on the sum of Observations , 1952 .

[3]  Solomon Kullback,et al.  Information Theory and Statistics , 1960 .

[4]  C. Radhakrishna Rao,et al.  Efficient Estimates and Optimum Inference Procedures in Large Samples , 1962 .

[5]  N. Chentsov.  A Systematic Theory of Exponential Families of Probability Distributions , 1966 .

[6]  J. Pfanzagl.  Asymptotic Expansions Related to Minimum Contrast Estimators , 1973 .

[7]  H. Akaike.  A new look at the statistical model identification , 1974 .

[8]  B. Efron.  Defining the Curvature of a Statistical Problem (with Applications to Second Order Efficiency) , 1975 .

[9]  A. Dawid.  Further Comments on Some Comments on a Paper by Bradley Efron , 1977 .

[10]  B. Efron.  THE GEOMETRY OF EXPONENTIAL FAMILIES , 1978 .

[11]  B. Efron,et al.  Assessing the accuracy of the maximum likelihood estimator: Observed versus expected Fisher information , 1978 .

[12]  J. F. C. Kingman,et al.  Information and Exponential Families in Statistical Theory , 1980 .

[13]  Rs Ingarden,et al.  INFORMATION THERMODYNAMICS AND DIFFERENTIAL GEOMETRY , 1979 .

[14]  Yoshiharu Sato,et al.  The geometrical structure of the parameter space of the two-dimensional normal distribution , 1979 .

[15]  David Hinkley,et al.  Theory of Statistical Estimation: The 1925 Paper , 1980 .

[16]  K. Takeuchi,et al.  Asymptotic efficiency of statistical estimators : concepts and higher order asymptotic efficiency , 1981 .

[17]  C. Atkinson.  Rao's distance measure , 1981 .

[18]  R. Ingarden.  Information geometry in functional spaces of classical and quantum finite statistical systems , 1981 .

[19]  S. Amari.  Geometrical theory of asymptotic ancillarity and conditional inference , 1982 .

[20]  S. Amari,et al.  Differential geometry of edgeworth expansions in curved exponential family , 1983 .