Recent Advances in Particle Swarm Optimization


ABSTRACT

Since its inception in 1995, research in particle swarm optimization (PSO) has been very active, to produce a large number of new PSO algorithms to improve performance and to solve a wide range of  optimization problem types. A good number of research articles  have  also  been produced on theoretical analyses of PSO. While PSO research has been quite extensive, and some may argue that it has reach a level of saturation, research in PSO remains very active, and continues to produce new PSO approaches to solve complex optimization problems and to better understand the behavior of particle swarms. This keynote will provide a review of the most recent developments in PSO. The focus will be with reference to two major aspects: (1) advances in gaining a better understanding of PSO, and (2) recent developments in producing new, more advanced, and more efficient PSO algorithms. With reference to gaining a better understanding of PSO, a short discussion of recent theoretical analysis will be given, focusing on those analyses that provide guidance to how control parameters should be initialized for optimal performance. The bulk of the discussion will focus on new PSO algorithms, with specific reference to self-adaptive PSO, heterogeneous PSO, new findings with respect to solving large- dimensional optimization problems.

Click to view keynote's slide