Estimation of Distribution Algorithm

From AGIRI.org

EDA's are a kind of evolutionary algorithm in which there is no crossover and mutation, but instead new candidate solutions are iteratively generated via

building a probabilistic model of the differences between relatively fit and relatively unfit members of the population
generating new instances based on this probability distribution

The most famous EDA is BOA, the Bayesian Optimization Algorithm.

MOSES is an EDA specialized for solving hard program learning problems.

[edit]