Adaptive dropout for high-dimensional expensive multiobjective optimization

Abstract Various works have been proposed to solve expensive multiobjective optimization problems (EMOPs) using surrogate-assisted evolutionary algorithms (SAEAs) in recent decades. However, most existing methods focus on EMOPs with less than 30 decision variables, since a large number of training samples are required build an accurate surrogate model for high-dimensional EMOPs, which is unrealistic optimization. To address this issue, we propose SAEA adaptive dropout mechanism. Specifically, mechanism takes advantage the statistical differences between different solution sets space guide selection some crucial variables. A new infill criterion then optimize selected variables assistance models. Moreover, optimized extended full-length solutions, and candidate solutions evaluated functions update archive. The algorithm tested benchmark up 200 compared state-of-the-art SAEAs. experimental results demonstrated promising performance computational efficiency