Greedy Particle Swarm Optimization Approach Using Leaky ReLU Function for Minimum Spanning Tree Problem
Ashish Kumar Singh, Anoj Kumar · 2025
The minimum spanning tree (MST) problem, a well-known optimization challenge, finds practical use across diverse fields including network design, transportation, and logistics. This paper presents an innovative approach to tackle the MST problem, integrating Greedy Particle Swarm Optimization (GPSO) with the Leaky Rectified Linear Unit (ReLU) function. GPSO, inspired by the collective behaviors of birds and fish, excels in efficiently exploring solution spaces and locating optimal solutions, forming the core of our methodology. To augment its performance, we introduce the Leaky ReLU function as the activation function in the particle swarm optimization process. The Leaky ReLU function, a recently introduced activation function, demonstrates promising qualities for optimization tasks. Its integration into the GPSO framework enables our approach to simultaneously encourage exploration and exploitation, leading to a more efficient search for the minimum spanning tree. This study offers a comprehensive analysis of our GPSO with the Leaky ReLU approach, encompassing detailed algorithmic descriptions, parameter configurations, and rigorous experimental results conducted on randomly generated 20, 40, 60, 80 nodes within the search space and generating a minimum spanning tree corresponding to it. The experiments conclusively demonstrate that our approach surpasses traditional MST algorithms, achieving competitive results when compared to state-of-the-art methods. In conclusion, our proposed Greedy Particle Swarm Optimization approach, combined with the Leaky ReLU function, presents a promising solution to the minimum spanning tree problem. Leveraging the synergy between GPSO and the Leaky ReLU function, we contribute to the advancement of optimization techniques applicable to graph-based problems, with potential implications across a range of real-world scenarios. The findings of this research suggest that the proposed approach holds promise for addressing MST problems in practical applications, showcasing its potential as a valuable tool in various domains.