Objective space partitioning using conflict information for solving many-objective problems

We present an algorithm that partitions the objective space based on an analysis of the conflict information obtained from the current Pareto front approximation. By partitioning the objectives in terms of the conflict among them, we aim to separate the multiobjective optimization into several subproblems in such a way that each of them contains the information to preserve as much as possible the structure of the original problem. We implement this framework by performing ranking and parent selection independently in each subspace. Our experimental results show that the proposed conflict-based partition strategy outperforms a similar algorithm in a test problem with independent groups of objectives. In addition, the new strategy achieves a better convergence and distribution than that produced by a strategy that creates subspaces at random. In problems in which the degree of conflict among the objectives is significantly different, the conflict-based strategy presents a better performance.