Title :
Renyi information divergence via measure transformations on minimal spanning trees
Author :
Hero, Alfred O. ; Michel, Olivier J J
Author_Institution :
Dept. of Electr. Eng. & Comput. Sci., Michigan Univ., Ann Arbor, MI, USA
Abstract :
We apply the results of Basseville (1989) to estimation of Renyi I-divergence between an unknown distribution and a known reference distribution using power weighted pruned minimal graphs spanning a random sample of n points from the unknown distribution. In particular we establish that the weight of a minimal graph connecting the points converges a.s. in n to the I-divergence after a suitable change of measure
Keywords :
entropy; graph theory; random processes; signal sampling; Renyi information divergence; entropy estimation; known reference distribution; measure transformations; minimal graph weight; minimal spanning trees; power weighted pruned minimal graphs; random sample; unknown distribution; Entropy; Joining processes; Multidimensional signal processing; Multidimensional systems; Particle measurements; Pattern recognition; Robustness; Statistical distributions; Statistics; Tree graphs;
Conference_Titel :
Information Theory, 2000. Proceedings. IEEE International Symposium on
Conference_Location :
Sorrento
Print_ISBN :
0-7803-5857-0
DOI :
10.1109/ISIT.2000.866712
