
[full paper] 
Michael Defoin Platel, Manuel Clergue, Philippe Collard
Genetic Programming (GP) has shown to be a good method to predict functions that solve inverse problems. In this context, a solution given by GP generally consists in a sole predictor. In contrast, Stackbased GP systems manipulate structures that allows to evolve together several predictors, that can be considered as teams of predictors. Work in machine learning reports that combining predictors gives good result in both quality and robustness. In this paper, we use Stackbased GP to study different ways to realize cooperation between predictors. First, preliminary tests and parameters tuning are performed on an academic GP benchmark. Then, the system is applied to a real inverse problem. A comparative study with standard methods has shown limits and advantages of teams prediction; in particular, when combinations take into account the response quality of each team member.
Keywords: Genetic Algorithms (Genetic Programming), Learning
Citation: Michael Defoin Platel, Manuel Clergue, Philippe Collard: Teams of Genetic Predictors for Inverse Problem Solving. In R.López de Mántaras and L.Saitta (eds.): ECAI2004, Proceedings of the 16th European Conference on Artificial Intelligence, IOS Press, Amsterdam, 2004, pp.989990.