EVOLVE -- a bridge between probability, set oriented numerics and evolutionary computation. Selected papers based on the presentations at the workshop 2011, Bourglinster Castle, Luxembourg, May 25--27, 2011. (Q441164)

The bridge between probability, set oriented numerics and evolutionary computation is created using a selection of contributions (invited talks and best full length papers) that were presented at the international workshop EVOLVE2011, jointly organised by researchers from Luxembourg, Mexico and France. While having as start point heuristic evolutionary algorithms, which are currently extensively used to approximate complex optimisation problems, the volume is aimed more at establishing the theoretical background for the cutting edge techniques that are proposed. The volume is organised in four parts. The first part ``Foundations, Probability and Evolutionary Computation'' is built around the description of probabilistic driven paradigms. The first chapter covers background notions of evolutionary algorithms described from a probabilistic perspective. Approaches such as stochastic optimisation, signal processing and analysis of convergence are included. The second and third chapter continue along the same line with the discussion of estimation of distribution algorithms (EDAs). The use of regular vines within EDAs through copula functions and Gaussian poly-tree estimation on graph based models inside continuous EDAs are presented. The second part of the volume revolves around set oriented numerics focused on multi-objective optimization. This part commences with a review of distances and quality indicators (Chapter 4), it is continued with the recent advances in set oriented techniques, with an emphasis on Pareto set approximations and concludes with basic algorithms for multi-objective optimization presented through optimal control problems (Chapter 5). The third part of the book ``Landscape, Coevolution and Cooperation'' changes the focus to genetic algorithms. The evolutionary paradigm is linked to Bayesian networks structure learning through examples (Chapter 6). In Chapters 7 and 8 hybrid algorithms (such as the one modelling the ripening process of the Camembert Cheese) are presented. An outline of the current state of art for Particle Swarm Optimisation and cellular Genetic Algorithms is also included. The forth part ``Multi-objective optimization, heuristic conversion algorithms'' presents the multi-objective optimization from a theoretical perspective and underlines the major problems using real life applications. First objective-function gradients are used to construct local search operators and the method is directly compared with existing approaches (Chapter 9). Next, the transition between single objective scheduling to multi-objective optimization is presented and the separation of multi objective optimisers into a modular composition of single objective scheduling heuristics is discussed. In Chapter 11 the single objective problems are reformulated and solved in a multi-objective form. This part concludes with a practical example of applying evolutionary algorithms to the real-world stock market. Although written in accessible format for undergraduates and researchers alike, the volume is particularly aimed at junior researchers in search for challenging topics. The technical background offered can be used as starting point for doctoral and post-doctoral dissertations.