This special session will cover original and pioneering contributions, theory as well as applications on creating and combining learning models, and aim at an inspiring discussion on the recent progress and the future development. Learners based on different paradigms can be combined for improved accuracy. Each learning method assumes a certain model that comes with a set of assumptions which may lead to error if the assumptions do not hold. Learning is an ill-posed problem and with finite data each algorithm converges to a different solution and fails under variant circumstances. In learning models combination, it is possible to make a distinction between two main modes: ensemble and modular. For an ensemble approach, several solutions to the same task, or task component, are combined to yield a more reliable estimate. In the modular approach case, particular aspects of a task are deal with by specialist components before being recombined to form a global solution. In this special session, the reasons for combining learning models and the main methods for creating and combining will be presented. Also, the effectiveness of these methods will be discussed considering the concepts of diversity and selection of these approaches.

• Modular approaches
• Hybrides systems
• Collaboratif learning
• Mixtures of distributions
• Mixtures of experts
• Ensemble methods
• Bagging approaches
• Boosting techniques
• Task decomposition

Younès Bennani, Full Professor - Paris 13 University         http://www-lipn.univ-paris13.fr/~bennani/

Nistor Grozavu, Associate Professor - Paris 13 University         http://www-lipn.univ-paris13.fr/~grozavu/

Mohamed Nadif, Full Professor - Paris 5 University         http://www.math-info.univ-paris5.fr/~nadifmoh/

Nicoleta Rogovschi, Associate Professor - Paris 5 University        http://www-lipn.univ-paris13.fr/~rogovschi/