Constraint Optimization Techniques for Exact Multi-Objective Optimization

MultiObjective Branch and Bound search has not been widely studied in the multiobjective context. The main reason is the lack of general approximation algorithms to compute lower bound sets. However, many lower bound techniques has been proposed for mono-objective optimization in the constraint programming field. In particular, Mini-Bucket Elimination (MBE) is a powerful mechanism for lower bound computation. Recently, MBE has been extended to multi-objective optimization problems. The new algorithm, called MO-MBE, computes a lower bound set of the efficient frontier of the problem. We show how to embed MO-MBE in a multi-objective branch and bound search, and we empirically demonstrate the performance of the new approach in two different domains.