Bayesian Optimization Algorithms for Multi-Objective Optimization

probabilistic models,Estimation Distribution Algorithms,multi-objective evolutionary optimization, Pareto-optimal solutions,Bayesian Optimization Algorithm, binary decision trees, knapsackproblem.

In recent years, several researchers have concentrated on usingprobabilistic models in evolutionary algorithms. These EstimationDistribution Algorithms (EDA) incorporate methods for automatedlearning of correlations between variables of the encoded solutions.The process of sampling new individuals from a probabilistic modelrespects these mutual dependencies among genes such that disruption ofimportant building blocks is avoided, in comparison with classicalrecombination operators. The goal of this paper is to investigate theusefulness of this concept in multi-objective evolutionaryoptimization, where the aim is to approximate the set of Pareto-optimalsolutions. We integrate the model building and sampling techniques of aspecial EDA called Bayesian Optimization Algorithm based on binarydecision trees into a general evolutionary multi-objective optimizer. Apotential performance gain is empirically tested in comparison withother state-of-the-art multi-objective EA on the bi-objective 0/1knapsack problem.