Towards a New Evolutionary Computation: Advances on Estimation of Distribution Algorithms (Studies in Fuzziness and Soft Computing)

Estimation of Distribution Algorithms (EDAs) are a set of algorithms in the Evolutionary Computation (EC) field characterized by the use of explicit probability distributions in optimization. Contrarily to other EC techniques such as the broadly known Genetic Algorithms (GAs) in EDAs, the crossover and mutation operators are substituted by the sampling of a distribution previously learnt from the selected individuals. EDAs have experienced a high development that has transformed them into an established discipline within the EC field.

This book attracts the interest of new researchers in the EC field as well as in other optimization disciplines, and that it becomes a reference for all of us working on this topic. The twelve chapters of this book can be divided into those that endeavor to set a sound theoretical basis for EDAs, those that broaden the methodology of EDAs and finally those that have an applied objective.

This is a nicely edited volume on Estimation of Distribution Algorithms (EDAs) by leading researchers on this important topic.

It covers a wide range of topics in EDAs, from theoretical analysis to experimental studies, from single objective to multi-objective optimisation, and from parallel EDAs to hybrid EDAs. It is a very useful book for everyone who is interested in EDAs, evolutionary computation or optimisation in general.

Xin Yao, IEEE Fellow
Editor-in-Chief, IEEE Transactions on Evolutionary Computation

______________________________________________________________

Estimation of Distribution Algorithms (EDAs) have "removed genetics"

from Evolutionary Algorithms (EAs). However, both approaches (still) have a lot in common, and, for instance, each one could be argued to in fact include the other! Nevertheless, whereas some theoretical approaches that are specific to EDAs are being proposed, many practical issues are common to both fields, and, though proposed in the mid 90's only, EDAs are catching up fast now with EAs, following many research directions that have proved successful for the latter:

opening to different search domains, hybridizing with other methods (be they OR techniques or EAs themselves!), going parallel, tackling difficult application problems, and the like.

This book proposes an up-to-date snapshot of this rapidly moving field, and witnesses its maturity. It should hence be read ... rapidly, by anyone interested in either EDAs or EAs, or more generally in stochastic optimization.

Marc Schoenauer
Editor-in-Chief, Evolutionary Computation