Estimation of Distribution Algorithms

Abstract

A Vine Estimation of Distribution Algorithm (VEDA) is a recently proposed optimization procedure built on top of a probabilistic graphical model called vine. The first target of vines was uncertainty analysis with high dimensional dependence modeling. The aim of this communication is to draw a path through a simple set of experiments, from the Univariate Marginal Distribution Algorithm to VEDA. Four algorithms are investigate in relation to their ability to deal with both weak and strong correlated variables in continuous unconstrained optimization problems. The results show that the models complement each other, although VEDA is the most promising algorithm.