Bayesian preference elicitation for decision support in multiobjective optimization
This provides a decision support tool for users in optimization domains, but it is incremental as it builds on existing Bayesian and elicitation methods.
The paper tackles the problem of helping decision-makers identify preferred solutions in multiobjective optimization by using a Bayesian model to estimate utility functions from pairwise comparisons, demonstrating superior performance in finding high-utility solutions with a small number of queries in experiments with up to nine objectives.
We present a novel approach to help decision-makers efficiently identify preferred solutions from the Pareto set of a multi-objective optimization problem. Our method uses a Bayesian model to estimate the decision-maker's utility function based on pairwise comparisons. Aided by this model, a principled elicitation strategy selects queries interactively to balance exploration and exploitation, guiding the discovery of high-utility solutions. The approach is flexible: it can be used interactively or a posteriori after estimating the Pareto front through standard multi-objective optimization techniques. Additionally, at the end of the elicitation phase, it generates a reduced menu of high-quality solutions, simplifying the decision-making process. Through experiments on test problems with up to nine objectives, our method demonstrates superior performance in finding high-utility solutions with a small number of queries. We also provide an open-source implementation of our method to support its adoption by the broader community.