A Surrogate-Assisted Evolutionary Algorithm with Innovized Progress Operator for Expensive Many-Objective Optimization
Abstract
To deal with challenging expensive many-objective optimization problems, this study proposes a surrogate-assisted evolutionary algorithm integrating an innovized progress operator and a novel infill criterion. Instead of relying solely on conventional genetic variation operators, innovized progress operator performs direction-aware boundary and gap progression under reference-vector guidance, thereby generating promising candidate solutions with improved exploration efficiency. Moreover, we design potential-driven expected improvement (PDEI) to select solutions for expensive evaluation by jointly modeling convergence and diversity indicators, which provides informative and balanced sampling pressure under a limited evaluation budget. The proposed algorithm achieves competitive performance on benchmark test suites compared with state-of-the-art algorithms.
Authors: Xiaochun Yang, Wei Hua Fang, Shuwei Zhu, Kalyanmoy Deb, Meiji Cui
Published in: Genetic and Evolutionary Computation Conference Companion (GECCO Companion) (2026)