A hybrid multi-objective evolutionary algorithm for the uncapacitated exam proximity problem

A hybrid multi-objective evolutionary algorithm for the uncapacitated exam proximity problem

Côté, Pascal and Wong, Tony and Sabourin, Robert

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 2005

Abstract : A hybrid Multi-Objective Evolutionary Algorithm is used to tackle the uncapacitated exam proximity problem. In this hybridization, local search operators are used instead of the traditional genetic recombination operators. One of the search operators is designed to repair unfeasible timetables produced by the initialization procedure and the mutation operator. The other search operator implements a simplified Variable Neighborhood Descent meta-heuristic and its role is to improve the proximity cost. The resulting non dominated timetables are compared with those produced by other optimization methods using 15 public domain datasets. Without special fine-tuning, the hybrid algorithm was able to produce timetables with good rankings in nine of the 15 datasets. © Springer-Verlag Berlin Heidelberg 2005.