Special issue on learning and intelligent optimization

Learning and optimization are often treated as two different research fields. However, it is clear that human problem solving strongly relies on learning in various forms. For sure, learning takes place when facing an initially unknown problem. In addition, while trying to solve it, by learning one obtains insights into its structure and derives improved strategies for solving it. Clearly, previous experience with problem solving also helps us humans to solve newly arising problems with increased efficiency. When seen from this perspective, learning and solving optimization problems are two closely related fields. The LION conference series on Learning and Intelligent Optimization makes this link explicit by declaring it its main theme. This special issue collects the extended versions of papers that have been presented at the second and the third LION conferences and a number of additional papers that have been contributed by authors active in this interdisciplinary research subject. Of the sixteen submitted papers, six have been accepted for publication. The first paper by Olivier Caelen and Gianluca Bontempi on A Dynamic Programming Strategy to Balance Exploration and Exploitation in the Bandit Problem introduces a new algorithm that is based on dynamic programming and on estimation techniques to tackle the well-known K-armed bandit problem. The following article, Analyzing Bandit-based Adaptive Operator Selection Mechanisms by Alvaro Fialho, Luis Da Costa, Marc Schoenauer and Michele Sebag, studies how the multi-armed