Abstract
Model management plays an essential role in surrogate-assisted evolutionary optimization of expensive problems, since the strategy for selecting individuals for fitness evaluation using the real objective function has substantial influences on the final performance. Among many others, infill criterion driven Gaussian process assisted evolutionary algorithms have been demonstrated competitive for optimization of problems with up to 50 decision variables. In this paper, a multi-objective infill criterion that considers the approximated fitness and the approximation uncertainty as two objectives is proposed for a Gaussian process assisted social learning particle swarm optimization algorithm. The multi-objective infill criterion uses non-dominated sorting for model management, thereby avoiding combining the approximated fitness and the approximation uncertainty into a scalar function, which is shown to be particularly important for high-dimensional problems, where the estimated uncertainty becomes less reliable. Empirical studies on 50- and 100-dimensional benchmark problems and a synthetic problem constructed from four real-world optimization problems demonstrate that the proposed multi-objective infill criterion is more effective than existing scalar infill criteria for Gaussian process assisted optimization given a limited computational budget.