Incremental Particle Swarm-Guided Local Search for Continuous Optimization

Abstract

We present an algorithm that is inspired by theoretical and empirical results in social learning and swarm intelligence research. The algorithm is based on a framework that we call incremental social learning. In practical terms, the algorithm is a hybrid between a local search procedure and a particle swarm optimization algorithm with growing population size. The local search procedure provides rapid convergence to good solutions while the particle swarm algorithm enables a comprehensive exploration of the search space. We provide experimental evidence that shows that the algorithm can find good solutions very rapidly without compromising its global search capabilities.

An incremental particle swarm for large-scale continuous optimization problems: an example of tuning-in-the-loop (re)design of optimization algorithms

“…When IPSOLS was originally proposed (Montes de Oca et al 2008, we used Powell's conjugate directions method (Powell 1964). In this paper, we explore the impact of the local search method on the performance of IPSOLS.…”

“…In the incremental particle swarm optimizer with local search (IPSOLS) (Montes de Oca et al 2008, the population size grows over time. Instead of a parameter that determines the population size, IPSOLS has a parameter, k, which is referred to as growth period.…”

“…We see two advantages in such a tuning-in-the loop approach: (1) the parameter search space is kept to a reasonable size, and (2) the result of tuning can help the designer to gain insight into the interactions between algorithmic components and their behavior on the optimization problem at hand. As a case study, we chose to redesign IPSOLS (Montes de Oca et al 2008, a particle swarm optimization (PSO) algorithm (Kennedy and Eberhart 2001) in which the size of the swarm grows over time and the particles' personal best solutions go through a phase of improvement via a local search procedure. We describe the redesign process and study the final algorithm's performance scalability as the dimensionality of the problem increases.…”

Incremental Social Learning Applied to a Decentralized Decision-Making Mechanism: Collective Learning Made Faster

“…Thus, it is reasonable to apply ISL for enhancing the performance of population-based optimization algorithms. In fact, ISL has been applied to a family of swarmbased optimization algorithms, namely, the particle swarm optimization algorithm [14,25,8], and encouraging results have been obtained [21,20]. Regarding the decentralized decision-making mechanism described in Section 2, related work is the one based on the simulation of the pheromone-laying and pheromone-following behavior of some ant species [13].…”

Stochastic Model of Evolutionary and Immunological Multi-Agent Systems: Mutually Exclusive Actions

“…Some of these actions cover the classic genetic operators like mixing (genotype exchanging) or eliminating of weak solutions (selection-like activities), some others tend directly to collect and improve their knowledge about solution. Basic ideas of memetic, agent-based approach may be found in [10,19].…”