Hessian Structures and Divergence Functions on Deformed Exponential Families

A Hessian structure \((\nabla , h)\) on a manifold is a pair of a flat affine connection \(\nabla \) and a semi-Riemannian metric \(h\) which is given by a Hessian of some function. In information geometry, it is known that an exponential family naturally has dualistic Hessian structures and their canonical divergences coincide with the Kullback-Leibler divergences, which are also called the relative entropies. A deformed exponential family is a generalization of exponential families. A deformed exponential family naturally has two kinds of dualistic Hessian structures and conformal structures of Hessian metrics. In this paper, geometry of such Hessian structures and conformal structures are summarized. In addition, divergence functions on these Hessian manifolds are constructed from the viewpoint of estimating functions. As an application of such Hessian structures to statistics, a generalization of independence and geometry of generalized maximum likelihood method are studied.

[1]  Estimators, escort probabilities, and phi-exponential families in statistical physics , 2004, math-ph/0402005.

[2]  S. Eguchi Geometry of minimum contrast , 1992 .

[3]  Shun-ichi Amari,et al.  Geometry of q-Exponential Family of Probability Distributions , 2011, Entropy.

[4]  Hirohiko Shima,et al.  Geometry of Hessian Structures , 2013, GSI.

[5]  A. Ohara,et al.  Information geometry of q-Gaussian densities and behaviors of solutions to related diffusion equations , 2008, 0810.0624.

[6]  Hiroshi Matsuzoe,et al.  Hessian Structures on Deformed Exponential Families , 2013, GSI.

[7]  Takashi Kurose Conformal-Projective Geometry of Statistical Manifolds , 2002 .

[8]  A. Ohara Geometric study for the Legendre duality of generalized entropies and its application to the porous medium equation , 2009, 0904.1530.

[9]  K. Takeuchi,et al.  Conformal Geometry of Statistical Manifold with Application to Sequential Estimation , 2011, 1102.3471.

[10]  Calyampudi R. Rao,et al.  Chapter 4: Statistical Manifolds , 1987 .

[11]  Noboru Murata,et al.  A Generalization of Independence in Naive Bayes Model , 2010, IDEAL.

[12]  Peter E. Jupp,et al.  Statistics, yokes and symplectic geometry , 1997 .

[13]  H. Matsuzoe STATISTICAL MANIFOLDS AND GEOMETRY OF ESTIMATING FUNCTIONS , 2013 .

[14]  C. Tsallis Introduction to Nonextensive Statistical Mechanics: Approaching a Complex World , 2009 .

[15]  H. Matsuzoe Geometry of contrast functions and conformal geometry , 1999 .

[16]  Noboru Murata,et al.  A generalisation of independence in statistical models for categorical distribution , 2012, Int. J. Data Min. Model. Manag..

[17]  Shun-ichi Amari,et al.  CONFORMAL GEOMETRY OF ESCORT PROBABILITY AND ITS APPLICATIONS , 2012 .

[18]  Shun-ichi Amari,et al.  Geometry of deformed exponential families: Invariant, dually-flat and conformal geometries , 2012 .

[19]  C. Tsallis Possible generalization of Boltzmann-Gibbs statistics , 1988 .

[20]  S. Amari,et al.  Asymptotic theory of sequential estimation : Differential geometrical approach , 1991 .

[21]  Takafumi Kanamori,et al.  Information Geometry of U-Boost and Bregman Divergence , 2004, Neural Computation.

[22]  Ernesto P. Borges A possible deformed algebra and calculus inspired in nonextensive thermostatistics , 2003, cond-mat/0304545.

[23]  Hiroki Suyari,et al.  Law of error in Tsallis statistics , 2005, IEEE Transactions on Information Theory.

[24]  Rui F. Vigelis,et al.  On φ-Families of Probability Distributions , 2013 .

[25]  Takashi Kurose ON THE DIVERGENCES OF 1-CONFORMALLY FLAT STATISTICAL MANIFOLDS , 1994 .

[26]  A. Ohara,et al.  Geometry for q-exponential families , 2011 .

[27]  M. C. Jones,et al.  Robust and efficient estimation by minimising a density power divergence , 1998 .

[28]  Masaru Tanaka,et al.  Meaning of an escort distribution and τ-transformation , 2010 .

[29]  On dual-projectively flat affine connections , 1995 .

[30]  Hiroshi Matsuzoe,et al.  Computational Geometry from the Viewpoint of Affine Differential Geometry , 2009, ETVC.

[31]  Filtrations for which All ℋ2 Martingales Are of Integrable Variation; Distances between σ-Algebras , 2008 .

[32]  Asuka Takatsu,et al.  Behaviors of φ-Exponential Distributions in Wasserstein Geometry and an Evolution Equation , 2011, SIAM J. Math. Anal..