Title :
On minimum divergence adaptation of discrete bivariate distributions to given marginals
Author :
Vajda, Igor ; Van der Meulen, Edward C.
Author_Institution :
Inst. of Inf. Theor. & Autom., Acad. of Sci. of the Czech Republic, Prague, Czech Republic
Abstract :
This correspondence considers the class P of discrete bivariate distributions p with marginals a(p), b(p) and the subclass Pa,b ⊂ P of distributions with given marginals a and b. For an arbitrary q ∈ P we study the distributions p˜ ∈ Pa,b minimizing on Pa,b various φ-divergences Dφ(p, q) for convex φ with finite φ(0).
Keywords :
information theory; maximum likelihood estimation; probability; ψ-divergence; Hellinger-Pearson divergence; contingency table approximation; discrete bivariate distribution; information divergence; minimum divergence adaptation; Automation; Communication networks; Frequency estimation; Information theory; Mathematics; Maximum likelihood estimation; Stochastic processes; Voting;
Journal_Title :
Information Theory, IEEE Transactions on
DOI :
10.1109/TIT.2004.839523