Local Binary Pattern-Based Adaptive Differential Evolution for Multimodal Optimization Problems.

Area of Science:

• Computational Intelligence
• Optimization Algorithms

Background:

• Multimodal optimization problems (MMOPs) require finding multiple global optima simultaneously.
• Existing algorithms often struggle with efficiently locating and maintaining diverse solutions in MMOPs.

Purpose of the Study:

• To propose a novel differential evolution (DE) algorithm, LBPADE, for efficient MMOP solving.
• To leverage Local Binary Pattern (LBP) principles for niching and identifying multiple peak regions.

Main Methods:

• Developed an LBP-based adaptive DE (LBPADE) algorithm.
• Incorporated a niching and global interaction (NGI) mutation strategy.
• Implemented an adaptive parameter strategy (APS) guided by LBP information.

Main Results:

• LBPADE effectively forms multiple niches to locate multiple peak regions in MMOPs.
• The NGI mutation and APS strategies enhance exploration and guide individuals to promising areas.
• Experimental results demonstrate LBPADE's superior or competitive performance against state-of-the-art algorithms.

Conclusions:

• The proposed LBPADE algorithm offers an efficient and effective approach for solving MMOPs.
• The integration of LBP principles with DE enhances the ability to find multiple global optima.