Accelerated high-dimensional global optimization: A particle swarm optimizer incorporating homogeneous learning and autophagy mechanisms

The curse of dimensionality often results in either premature convergence or slow convergence in high-dimensional global optimization. This paper proposes the high-speed homogeneous learning-based particle swarm optimizer (HLPSO) to address these challenges. The algorithm is based on the observation that individuals with similar attitudes or beliefs tend to coexist harmoniously in human societies, unlike birds or fish that squeeze or clash in the same space. By incorporating the concept of homogeneity into the optimization process, the algorithm controls particle learning and generates distinct subpopulations for competition and learning. Furthermore, to accelerate the convergence rate, the algorithm employs an autophagy mechanism and dynamic subpopulation diversity, guiding the population towards global optimal solutions while reducing redundant fitness evaluations. The effectiveness of HLPSO is illustrated through its exceptional performance in benchmark test suites, including the IEEE congress on evolutionary computation (CEC) 2008, CEC 2010, and CEC 2013, in both 1000 and 2000 dimensions. These results not only highlight its superiority over several state-of-the-art algorithms but also underscore its aptitude for effectively addressing low-dimensional optimization challenges encountered in CEC 2017.

Automated summary

This paper introduces a high-speed homogeneous learning-based particle swarm optimizer (HLPSO) to address the challenges of premature or slow convergence in high-dimensional global optimization. The algorithm incorporates the concept of homogeneity to control the optimization process and achieve faster convergence. Experimental results demonstrate the superior performance of HLPSO in benchmark test suites, indicating its aptitude for effectively addressing low-dimensional optimization challenges encountered in CEC 2017.