Abstract

Many real-life decision problems require managing trade-offs between multiple objectives. Evolutionary algorithms have been successfully applied to approximate the Pareto frontier and find a finite population of optimal solutions. But instead of a computation of the whole front, the reference point approaches can aggregate different strategies to drive the search on relevant areas previously selected by the decision maker. Additionally, the usage of collective preferences to build reference points enhances the multi-objective results. This work applies a collective intelligence MOEA to solve a continuous problem of facility location. The experimental results validate the concepts being proposed and demonstrate that the COIN algorithm requires fewer number of function evaluations for complex scenarios in comparison with state-of-the-art alternatives.