We have previously shown that a genetic algorithm can be approximated by an
evolutionary algorithm using the product of univariate marginal distributi
ons of selected points as search distribution. This algorithm (UMDA) succes
sfully optimizes difficult multi-modal optimization problems. For correlate
d fitness landscapes snore complex factorizations of the search distributio
n have to be used. These factorizations are used by the Factorized Distribu
tion Algorithm FDA. In this paper we extend FDA to an algorithm which compu
tes a factorization from the data. The factorization can be represented by
a Bayesian network. The Bayesian network is used to generate the search poi
nts.