Multi-granularity competition-cooperation optimization algorithm with adaptive parameter configuration

doi:10.1007/s10489-021-02952-9

Multi-granularity competition-cooperation optimization algorithm with adaptive parameter configuration

2022-02-23

2

PRE

AI

DOI：10.1007/s10489-021-02952-9

原文链接

原文求助

摘要

En 中文

The intelligent optimization algorithm has the advantage of giving feasible solutions in polynomial time when solving complex problems in reality. Its performance depends on its strategy design and parameter configuration. The key to achieving computational intelligence and self-organization is parameter configuration. However, the parameter configuration in the intelligent optimization algorithm relies on the designer's experience, and the parameter adjustment process will waste a vast amount of resources. In terms of this issue, this paper proposes a multi-granularity competition-cooperation optimization algorithm with adaptive parameter configuration(MGAP). First, the particles are grouped into groups according to the fitness of the solution. Second, an inter-group competition-cooperation relationship network is established. According to the relationship network, each particle performs adaptive learning between and within the group and updates. Then, reinforcement learning is introduced to train learning rules and learning parameters of adaptive learning between and within groups to promote particle autonomous evolution. We have done comparative experiments with six algorithms on the 10-100 dimension test functions and then give application cases. The comparison results with other algorithms in the optimization function show that the proposed algorithm is computationally effective and efficient in the face of large-scale optimization problems. Further experiments on complex problems in the real scene of 270,000 campus consumption data verify the collaborative optimization performance and collaborative learning performance of our algorithm.

Keyword:

Group competition-cooperation

Reinforcement learning

Collaborative optimization

Large-scale optimization

Multi-granularity competition-cooperation optimization algorithm with adaptive parameter configuration

摘要

期刊

机构

引用论文