Open Access

ARTICLE

Adaptive Evolution of Metaheuristic Update Strategies Using Genetic Programming for Remote Sensing Image Fusion

Jeng-Shyang Pan^1,2,3, Wenda Li², Shu-Chuan Chu^4,*, Zhi-Gang Du⁵, Hongmei Yang², Lingping Kong⁶

1 School of Information Engineering, Yango University, Fuzhou, China
2 College of Computer Science and Engineering, Shandong University of Science and Technology, Qingdao, China
3 Department of Information Management, Chaoyang University of Technology, Taichung City, Taiwan
4 School of Artificial Intelligence, Nanjing University of Information Science and Technology, Nanjing, China
5 School of Transportation and Logistics, Southwest Jiaotong University, Chengdu, China
6 Faculty of Electrical Engineering and Computer Science, VSB-Technical University of Ostrava, Ostrava, Czech Republic

* Corresponding Author: Shu-Chuan Chu. Email:
# Presented at the International Conference on Machine Intelligence Theory and Applications 2026; 2026 Feb 24–28; Dunedin, New Zealand

(This article belongs to the Special Issue: Advances in Computational Intelligence for Complex Systems)

Computer Modeling in Engineering & Sciences 2026, 148(2), 33 https://doi.org/10.32604/cmes.2026.084683

Received 27 April 2026; Accepted 29 July 2026; Issue published 28 August 2026

Abstract

Formulating efficient updating techniques is essential for the efficacy of metaheuristic algorithms. Traditional approaches, however, depend significantly on manually developed formulas and empirical intuition, which frequently constrain their adaptability and scalability across various optimization tasks. This research introduces a Genetic Programming-based Metaheuristic framework, referred to as GP-MAs, designed to autonomously develop and enhance symbolic update rules for metaheuristic algorithms. Within the suggested GP-MAs architecture, genetic programming (GP) is integrated into the learning phase of the Growth Optimizer (GO) to dynamically formulate symbolic update equations, hence enhancing the algorithm’s adaptability to diverse optimization landscapes. A hybrid algorithm, termed GPbasedGO, is constructed based on GO to validate the effectiveness of the proposed framework. Comprehensive experiments on the CEC2022 benchmark suite illustrate the competitive convergence characteristics, optimization efficacy, and competitive performance across benchmark problems of GPbasedGO in comparison to many prominent state-of-the-art techniques. The practical usability of the proposed approach is assessed in a multispectral and panchromatic image fusion challenge, where GPbasedGO attains competitive results across many image quality measures, including ERGAS, SAM, RMSE, UIQI, and CC. The findings demonstrate that the GP-MAs framework offers a versatile and semi-automated model for metaheuristic optimization, proficient in producing adaptive symbolic updating techniques for key optimization problems. This study presents a comprehensive validation of GP-assisted symbolic strategy evolution for metaheuristic optimization and suggests interesting avenues for further research in adaptive optimization.

---

Keywords

---