Riemannian geometry on non-parametric probability space

被引:0
|
作者
Combe-Nencka, H.
Combe, Ph.
机构
[1] CNRS Lunimy, CPT, F-13288 Marseille 09, France
[2] Univ Aix Marseille 1, F-13331 Marseille 3, France
关键词
non-parametric statistics; probability spaces; information theory; real Hilbert spaces; Riemannian infinite manifolds; geodesics flows;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
The family P-lambda of absolutely continuous probabilities w.r.t. the sigma-finite measure lambda is equipped with a structure of an infinite dimensional Riemannian manifold modeled on a real Hilbert. Firstly, the relation between the Hellinger distance and the Fischer metric is analysed on the positive cone M-lambda(+) of bounded measures absolutely continuous w.r.t. lambda, appearing as a flat Riemannian manifold. Secondly, the statistical manifold P-lambda is seen as a submanifold of M-lambda(+) and Amari-Chensov alpha-connections are derived. Some alpha-self-parallel curves are explicitely exhibited.
引用
收藏
页码:459 / 473
页数:15
相关论文
共 50 条