Multiobjective optimization by decision diagrams

Abstract

In this paper we present a technique for solving multiobjective discrete optimization problems using decision diagrams. The proposed methodology is related to an algorithm designed for multiobjective optimization for dynamic programming, except utilizing decision diagram theory to reduce the state space, which can lead to orders of magnitude performance gains over existing algorithms. The decision diagram-based technique is applied to knapsack, set covering, and set partitioning problems, exhibiting improvements over state-of-the-art general-purpose multiobjective optimization algorithms.