Guest editorial: Special issue based on the LION 4 conference

The series of LION conferences (LION stands for Learning and Intelligent Optimization) aims at exploring the boundaries and uncharted territories between machine learning, artificial intelligence, mathematical programming and algorithms for hard optimization problems. The main purpose of these events is to bring together experts from the above mentioned areas to discuss new ideas and methods, challenges and opportunities in various application areas, general trends and specific developments. Following the tradition of earlier LION conferences, this special issue is dedicated to extended versions of four carefully selected papers presented at LION 4, which was held in Venice, Italy, in January of 2010. The first paper by Matteo Gagliolo and Jurgen Schmidhuber on Algorithm Portfolio Selection as a Bandit Problem with Unbounded Losses proposes a method that learns to allocate computation time to a given set of algorithms, of unknown performance, with the aim of solving a given sequence of problem instances in a minimum time. The second article, Discovering the Suitability of Optimisation Algorithms by Learning from Evolved Instances by Kate Smith-Miles and Jano van Hemert deals with performance prediction of algorithms from a portfolio. They propose the use of an evolutionary algorithm to evolve instances that are uniquely easy or hard for each algorithm from the portfolio, thus providing a direct method for studying the relative strengths and weaknesses of each algorithm. A case study is presented based