Combining Lipschitz and RBF surrogate models for high-dimensional computationally expensive problems

Standard evolutionary optimization algorithms assume that the evaluation of objective and constraint functions is straightforward computationally cheap. However, in many real-world problems, these evaluations involve expensive numerical simulations or physical experiments. Surrogate-assisted (SAEAs) have recently gained increased attention for their performance solving types problems. The main idea SAEAs integration an algorithm with a selected surrogate model approximates function. In this paper, we propose based on Lipschitz underestimation use it to develop differential evolution-based algorithm. algorithm, called Differential Evolution (LSADE), utilizes Lipschitz-based model, along standard radial basis function local search procedure. experimental results seven benchmark dimensions 30, 50, 100, 200 show proposed LSADE competitive compared state-of-the-art under limited computational budget, being especially effective very complicated high dimensions.