Abstract

Self-organization is a natural phenomenon that emerges in systems with alarge number of interacting components. Self-organized systems show robustness,scalability, and flexibility, which are essential properties when handlingreal-world problems. Swarm intelligence seeks to design nature-inspiredalgorithms with a high degree of self-organization. Yet, we do not know whyswarm-based algorithms work well and neither we can compare the differentapproaches in the literature. The lack of a common framework capable ofcharacterizing these several swarm-based algorithms, transcending theirparticularities, has led to a stream of publications inspired by differentaspects of nature without much regard as to whether they are similar to alreadyexisting approaches. We address this gap by introducing a network-basedframework$-$the interaction network$-$to examine computational swarm-basedsystems via the optics of social dynamics. We discuss the social dimension ofseveral swarm classes and provide a case study of the Particle SwarmOptimization. The interaction network enables a better understanding of theplethora of approaches currently available by looking at them from a generalperspective focusing on the structure of the social interactions.

 

Introduction (beta)

None

 

Conclusion (beta)

None