[Paper Review] Egret Swarm Optimization Algorithm: An Evolutionary Computation Approach for Model Free Optimization

Zuyan Chen, Adam Francis+3 more|arXiv (Cornell University)|Jul 29, 2022

Metaheuristic Optimization Algorithms Research4 citations

TL;DR

This paper proposes the Egret Swarm Optimization Algorithm (ESOA), a novel meta-heuristic optimization method inspired by the cooperative hunting behaviors of Great Egrets and Snowy Egrets. ESOA integrates a Sit-And-Wait Strategy, Aggressive Strategy, and Discriminant Conditions to balance exploration and exploitation, demonstrating superior performance and robustness across 36 benchmark functions and two real-world engineering problems compared to PSO, GA, DE, GWO, and HHO.

ABSTRACT

A novel meta-heuristic algorithm, Egret Swarm Optimization Algorithm (ESOA), is proposed in this paper, which is inspired by two egret species' (Great Egret and Snowy Egret) hunting behavior. ESOA consists of three primary components: Sit-And-Wait Strategy, Aggressive Strategy as well as Discriminant Conditions. The performance of ESOA on 36 benchmark functions as well as 2 engineering problems are compared with Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Differential Evolution (DE), Grey Wolf Optimizer (GWO), and Harris Hawks Optimization (HHO). The result proves the superior effectiveness and robustness of ESOA. The source code used in this work can be retrieved from https://github.com/Knightsll/Egret_Swarm_Optimization_Algorithm; https://ww2.mathworks.cn/matlabcentral/fileexchange/115595-egret-swarm-optimization-algorithm-esoa.

Motivation & Objective

• To address the limitations of existing meta-heuristic algorithms in balancing exploration and exploitation for non-convex, gradient-free optimization problems.
• To develop a novel nature-inspired algorithm that mimics the cooperative and adaptive hunting strategies of Great Egrets and Snowy Egrets.
• To evaluate ESOA’s performance on standard benchmark functions and real-world engineering optimization problems.
• To demonstrate ESOA’s superiority in convergence speed, solution quality, and stability compared to state-of-the-art algorithms.
• To provide open-source implementations to support broader application and further research in model-free optimization.

Proposed method

• ESOA models the Snowy Egret’s Sit-And-Wait Strategy, where agents remain stationary to detect and capture prey, representing a local search or exploitation phase.
• ESOA incorporates the Great Egret’s Aggressive Strategy, involving active movement and pursuit of prey, symbolizing global exploration and population diversification.
• A Discriminant Condition dynamically determines whether an agent should switch between Sit-And-Wait and Aggressive strategies based on fitness improvement and convergence criteria.
• The algorithm uses a population of search agents that update their positions iteratively using velocity and position update rules derived from the behavioral models.
• The framework integrates constraint handling via penalty functions, with the penalty coefficient φ set to 10^100 for the Belegundu problem and 10^5 for the spring design problem.
• The algorithm is evaluated using 36 CEC1997 and CEC2017 benchmark functions and two engineering problems: the Belegundu problem and the tension/compression spring design.

Experimental results

Research questions

• RQ1 Can a novel meta-heuristic algorithm inspired by egret hunting behavior achieve better balance between exploration and exploitation than existing methods?
• RQ2 How does ESOA perform on a diverse set of benchmark functions, including unimodal, multimodal, hybrid, and composition functions?
• RQ3 What is the convergence behavior and stability of ESOA across multiple independent runs on complex optimization problems?
• RQ4 Can ESOA effectively solve real-world constrained engineering optimization problems with high accuracy and robustness?
• RQ5 How does ESOA compare in performance and computational efficiency to PSO, GA, DE, GWO, and HHO on both synthetic and practical problems?

Key findings

• On the CEC1997 and CEC2017 benchmark sets, ESOA achieved the best mean performance on 24 out of 36 functions, demonstrating superior global search capability.
• For the Belegundu problem, ESOA achieved the best optimal result of -30665.18896, outperforming PSO (-30663.21135) and other algorithms in both best and mean values.
• In the tension/compression spring design problem, ESOA achieved the best average fitness of 0.012855, with the lowest standard deviation (7.34×10^-5), indicating high stability.
• ESOA showed the second-lowest standard deviation among all algorithms on the Belegundu problem, confirming its robustness despite not achieving the absolute best mean.
• The algorithm’s convergence behavior was consistently faster and more stable than PSO, GA, DE, GWO, and HHO across all test functions.
• The source code for ESOA is publicly available on GitHub and MathWorks File Exchange, enabling reproducibility and further research in model-free optimization.