Multi-objective multi-task optimization using decomposition collaboration mechanism and adaptive differential evolution

Jiani Fu

doi:10.1016/j.future.2026.108781

Multi-objective multi-task optimization using decomposition collaboration mechanism and adaptive differential evolution

2026-09-06

0

PRE

AI

DOI：10.1016/j.future.2026.108781

Original

Original request for help

Save

Abstract

En 中文

Multi-objective multi-task optimization (MOMTO) is a key research direction in evolutionary computation, widely used to address complex trade-offs among multiple objectives and tasks. In recent years, numerous algorithms have been proposed and demonstrated promising performance. However, with the increasing scale and complexity of real-world problems, especially in high-dimensional settings, existing methods face major challenges such as slow convergence, high computational cost, and loss of population diversity.To address these challenges, this paper proposes an innovative many-objective and multi-task optimization algorithm (MaMTO/D-ADE). The algorithm incorporates an efficient neighborhood cooperation mechanism based on a decomposition strategy to mitigate the high complexity and inefficiency of solving large-scale optimization problems. Furthermore, an enhanced differential evolution approach is developed, which integrates adaptive parameter control with hybrid mutation operators to effectively balance global exploration and local exploitation. A flexible offspring updating mechanism is also introduced to accelerate convergence while maintaining population diversity.Extensive experiments conducted on several well-established benchmark problems demonstrate that MaMTO/D-ADE achieves significant improvements in both convergence speed and solution quality, exhibiting superior competitiveness compared to existing state-of-the-art algorithms.

Multi-objective multi-task optimization using decomposition collaboration mechanism and adaptive differential evolution

Abstract

Journal

Organization

Cited Papers