Abstract:Many-objective optimization presents unique challenges in balancing diversity and convergence of solutions. Traditional approaches struggle with this balance, leading to suboptimal solution distributions in the objective space especially at higher number of objectives. This necessitates the need for innovative strategies to adeptly manage these complexities. This study introduces a Many-Objective Artificial Hummingbird Algorithm (MaOAHA), an advanced evolutionary algorithm designed to overcome the limitations of existing many-objective optimization methods. The objectives are to improve convergence rates, maintain solution diversity and achieve a uniform distribution in the objective space. MaOAHA implements information feedback mechanism (IFM), reference point-based selection and association, non-dominated sorting and niche preservation. The IFM utilizes historical data from previous generations to inform the update process, thereby improving the algorithm's the exploration and exploitation capabilities. Reference point-based selection, along with non-dominated sorting, ensures solutions are both close to the Pareto front and evenly spread in the objective space. Niche preservation and density estimation strategies are employed to maintain diversity and prevent overcrowding. The comprehensive experimental analysis benchmarks MaOAHA against four leading algorithms viz. MaOGBO, MaOPSO, RVEA and NSGA-III. The DTLZ1-DTLZ7 benchmark sets with 4, 6 and 8 objectives and five real-world problems (RWMaOP1-RWMaOP5) are considered for performance assessment of the selected algorithms. The results demonstrate that internal parameter free MaOAHA significantly outperforms its counterparts, achieving better GD by up to 52.38%, IGD by up to 38.09%, SP by up to 56%, SD by up to 71.42%, HV by up to 44% and RT by up to 52%. These metrics affirm the MaOAHA's capability to enhance the decision-making processes through its adept balance of convergence, diversity and uniformity.

Machine Learning Based Decision Support for Many-Objective Optimization Problems

Multimodal Multiobjective Differential Evolution Algorithm Based on Enhanced Decision Space Search

DMaOEA-εC: Decomposition-based many-objective evolutionary algorithm with the ε-constraint framework

Multi-objective evolution strategy for multimodal multi-objective optimization

Knee based multimodal multi-objective evolutionary algorithm for decision making

An Incremental Learning Evolutionary Algorithm for Many-Objective Optimization with Irregular Pareto Fronts

Machine learning aided multi-objective optimization and multi-criteria decision making: Framework and two applications in chemical engineering

Interactive Evolutionary Multi-Objective Optimization via Learning-to-Rank

A Projection-Based Evolutionary Algorithm for Multi-Objective and Many-Objective Optimization

A coevolutionary algorithm using multi-operator ensemble for many-objective optimisation problems

A reinforcement learning approach for dynamic multi-objective optimization

Multi‐Objective Evolutionary Algorithm Based on Decomposition With Orthogonal Experimental Design

Comparison of MOEAs in an Optimization-Decision Methodology for a Joint Order Batching and Picking System

A multistage evolutionary algorithm for many-objective optimization

Icb-Moea/D: An Interactive Classification-Based Multi-Objective Optimization Algorithm

A many-objective evolutionary algorithm based on learning assessment and mapping guidance of historical superior information

A New Hyper-Heuristic Multi-Objective Optimisation Approach Based on MOEA/D Framework

Cooperative co-evolutionary algorithm for multi-objective optimization problems with changing decision variables

Many-Objective Artificial Hummingbird Algorithm (MaOAHA): An Effective Many-Objective Algorithm for Engineering Design Problems

MOEA/D: A Multiobjective Evolutionary Algorithm Based on Decomposition

Multi-objective multi-criteria evolutionary algorithm for multi-objective multi-task optimization