Optimality conditions for maximizers of the information divergence from an exponential family

František Matúš

Displaying similar documents to “Optimality conditions for maximizers of the information divergence from an exponential family”

Comparison of two methods for approximation of probability distributions with prescribed marginals

Albert Pérez, Milan Studený (2007)

Kybernetika

Similarity:

Let $P$ be a discrete multidimensional probability distribution over a finite set of variables $N$ which is only partially specified by the requirement that it has prescribed given marginals ${P_{A}; A \in 𝒮}$ , where $𝒮$ is a class of subsets of $N$ with $⋃ 𝒮 = N$ . The paper deals with the problem of approximating $P$ on the basis of those given marginals. The divergence of an approximation $\hat{P}$ from $P$ is measured by the relative entropy $H (P | \hat{P})$ . Two methods for approximating $P$ are compared. One of them uses formerly introduced...

Distributivity of strong implications over conjunctive and disjunctive uninorms

Daniel Ruiz-Aguilera, Joan Torrens (2006)

Kybernetika

Similarity:

This paper deals with implications defined from disjunctive uninorms $U$ by the expression $I (x, y) = U (N (x), y)$ where $N$ is a strong negation. The main goal is to solve the functional equation derived from the distributivity condition of these implications over conjunctive and disjunctive uninorms. Special cases are considered when the conjunctive and disjunctive uninorm are a $t$ -norm or a $t$ -conorm respectively. The obtained results show a lot of new solutions generalyzing those obtained in previous works...

On Ozeki’s inequality for power sums

Horst Alzer (2000)

Czechoslovak Mathematical Journal

Similarity:

Let $p \in (0, 1)$ be a real number and let $n \geq 2$ be an even integer. We determine the largest value $c_{n} (p)$ such that the inequality $\sum_{i = 1}^{n} {| a_{i} |}^{p} \geq c_{n} (p)$ holds for all real numbers $a_{1}, ..., a_{n}$ which are pairwise distinct and satisfy ${min}_{i \neq j} | a_{i} - a_{j} | = 1$ . Our theorem completes results of Ozeki, Mitrinović-Kalajdžić, and Russell, who found the optimal value $c_{n} (p)$ in the case $p > 0$ and $n$ odd, and in the case $p \geq 1$ and $n$ even.

On the structure of continuous uninorms

Paweł Drygaś (2007)

Kybernetika

Similarity:

Uninorms were introduced by Yager and Rybalov [13] as a generalization of triangular norms and conorms. We ask about properties of increasing, associative, continuous binary operation $U$ in the unit interval with the neutral element $e \in [0, 1]$ . If operation $U$ is continuous, then $e = 0$ or $e = 1$ . So, we consider operations which are continuous in the open unit square. As a result every associative, increasing binary operation with the neutral element $e \in (0, 1)$ , which is continuous in the open unit square may be...

Complex series and connected sets

B. Jasek

Similarity:

CONTENTSPREFACE..........................................................................................................................................................................3INTRODUCTION............................................................................................................................................................. 41. Notation. 2. Subject of the paper.Chapter I. DECOMPOSITION OF Σ INTO $Σ_{1}$ , $Σ_{2}$ , $Σ_{3}$ , $Σ_{4}$ INESSENTIAL RESTRICTIONOF GENERALITY ...............................................................................................................................................................