Improved Swarm Algorithm Targets Convergence and Solution Spread in Benchmarks

Researchers have reported an improved multi-objective particle swarm optimization algorithm, called MCMOPSO, that integrates a density-morphology archive maintenance method with dynamic tracking matrix parameter adjustment. The work, published in the International Journal of Machine Learning and Cybernetics, addresses limitations of traditional multi-objective particle swarm optimization in achieving coordinated optimization of convergence efficiency and the distribution of the solution set.

The algorithm’s innovations are concentrated in three areas. The first is an external density-morphology archive maintenance scheme that achieves precise screening of non-dominated solutions. The second is a dynamic tracking matrix parameter adjustment that finely regulates particle trajectories. The third is a percentile position advantage strategy designed to efficiently guide the population toward the Pareto optimal frontier.

To evaluate the approach, the authors conducted comparative experiments between MCMOPSO and 10 classical algorithms on three benchmark test problem families: ZDT, UF, and DTLZ. Performance was quantitatively analyzed using metrics including hyper-volume (HV) and inverse generational distance (IGD).

The reported results show that MCMOPSO exhibits superior convergence, distribution uniformity, and stability across most of the test problems examined. According to the authors, the algorithm provides an efficient and reliable solution for complex multi-objective optimization.

The findings come from standardized benchmark problems rather than specific real-world deployments, so the reported advantages describe performance on those test suites. Within that scope, the study presents MCMOPSO as an alternative among multi-objective particle swarm methods, combining archive maintenance, trajectory adjustment, and position-based guidance to coordinate convergence and distribution.

Subject of Research: Multi-objective particle swarm algorithm based on density-morphology archive maintenance and dynamic tracking matrix parameter adjustment

Article Title: Multi-objective particle swarm algorithm based on density-morphology archive maintenance and dynamic tracking matrix parameter adjustment

Article References: Xu, J., Liu, Y., Chen, S., Jiang, Y., Yang, J., & Wang, Q. (2026). Multi-objective particle swarm algorithm based on density-morphology archive maintenance and dynamic tracking matrix parameter adjustment. International Journal of Machine Learning and Cybernetics, 17(11), Article 510. https://doi.org/10.1007/s13042-026-03340-w

Image Credits: AI Generated

DOI: 10.1007/s13042-026-03340-w

Keywords: Multi-objective, particle, swarm, algorithm, based, density-morphology, archive, maintenance, dynamic, tracking, matrix, parameter