Guest editorial: Special issue—revised selected papers of the LION 5 conference

The International Conference on Learning and IntelligentOptimizatioN (LION) has as its goal to explore the intersections between machine learning, artificial intelligence, mathematical programming and algorithms for hard optimization problems. The main purpose of this conference is to bring together experts from these disciplines, in order to discuss new ideas andmethods, as well as to identify challenges and opportunities in various application areas, general trends and specific developments. This special issue contains the extended and revised versions of three carefully selected papers that were presented at LION 5, which took place in Rome, Italy, during January 17–21, 2011. The first paper, from Sato, Aguirre and Tanaka, presents a study of the effectiveness of crossover and mutation operators in 0/1 knapsack problems having four or more objectives and up to 20 items. The authors show that recombining very dissimilar individuals, without setting a limit to the information being crossed, causes the operator to become too disruptive and, consequently, decreases its effectiveness. In order to deal with this problem, the authors propose to use a local recombination operator that selects mating parents based on proximity in objective function space as well as two-point and uniform crossover operators that control the maximumnumber of crossed genes. These operators are shown, through a series of experiments, to significantly improve performance. The second paper, from Kampouridis, Alsheddy and Tsang, presents a thorough study of different hyper-heuristics frameworks, with the aim of improving the performance of a financial forecasting tool, based on genetic programming, called EDDIE 8. The authors devised 14 heuristics, which were applied to 30 different