Information Manifold and Ricci Curvature

摘要

This paper discusses affine immersion of multivariable Gaussian statistical manifolds for multi-sensor networks. Firstly, the potential function that can represent the manifold is obtained from the statistical significance, that is, the shape of the manifold is expressed by the potential function, and the high-dimensional manifold of the sensor network information is embedded into the Euclidean space for research and representation, and its characteristics are studied. Then the Ricci curvature of the information space is calculated to obtain the degree of curvature of the manifold to represent the changing trend of information.