Marian Grendár

Research interests

Pi problem and Maximum Probability (MaxProb) method.
Relative Entropy Maximization (REM/MaxEnt) as an asymptotic instance of MaxProb.
Probabilistic justification of MaxProb and REM/MaxEnt via the  Conditional Law of Large Numbers.
Parametric and empirical extensions of REM/MaxEnt.

Phi problem and Bayesian Maximum Probability (MAP) method.
Maximum Non-parametric Likelihood (MNPL) as an asymptotic instance of MAP.
Probabilistic justification of MAP and MNPL via the Bayesian Law of Large Numbers.
Large-deviations approach to the Bayesian non-parametric consistency.
Estimating Equations and Empirical Likelihood.

A brief survey of the Pi and Phi problems can be found here.

Statistical evidence: likelihood

Misc: Jeffreys entropy maximization, Golan, Judge & Miller's Generalized MaxEnt, Maximum Entropy Production, Graph entropy.

Applications:  spam filtering, recurrence quantification analysis

                 list of selected works ¤ MaxProb site ¤ George Judge ¤ Robert K. Niven ¤ Arthur Ramer ¤ Laura Schechter ¤ Vladimír Špitalský ¤ iškolában