Bayesian sufficient statistics and invariance
Bemerkungen zum Vergleich linearer normaler Experimente.
Blended -divergences with examples
Several new examples of divergences emerged in the recent literature called blended divergences. Mostly these examples are constructed by the modification or parametrization of the old well-known phi-divergences. Newly introduced parameter is often called blending parameter. In this paper we present compact theory of blended divergences which provides us with a generally applicable method for finding new classes of divergences containing any two divergences and given in advance. Several examples...
Bound on extended -divergences for a variety of classes
The concept of -divergences was introduced by Csiszár in 1963 as measures of the ‘hardness’ of a testing problem depending on a convex real valued function on the interval . The choice of this parameter can be adjusted so as to match the needs for specific applications. The definition and some of the most basic properties of -divergences are given and the class of -divergences is presented. Ostrowski’s inequality and a Trapezoid inequality are utilized in order to prove bounds for an extension...
Boundedly Complete Families Which are Not Complete.
Bounds on margin distributions in learning problems