The Firefly Algorithm with Gaussian Disturbance and Local Search
Li Lv,Jia Zhao
DOI: https://doi.org/10.1007/s11265-017-1278-y
2017-09-06
Journal of Signal Processing Systems
Abstract:Along with the rapid development of mobile Internet, Internet of things and cloud computing technology, the data volume has shown an explosive growth in different industries. Big data technology, which provides new solutions to data-related problems, draws an increasing attention, especially in the field of artificial intelligence. Swarm intelligence is an important tool for solving complex problems in both scientific research and engineering practice. Representing a major development trend in artificial intelligence and information science, swarm intelligence has displayed great application potentials in big data analysis and data mining. Firefly algorithm (FA), an optimization technique based on swarm intelligence, has been successfully applied to a diversity of complex engineering optimization problems. In a standard FA, particles migrate blindly towards those better ones, without considering the status of the object of learning. However, this type of particle regeneration may result in a solution being trapped into local optima, with fast convergence speed but low convergence precision. We propose an FA with Gaussian disturbance and local search. The swarm is updated using random attraction model. The current position of the particle is compared with particle’s historical optimal position. If the current position is inferior to the historical optimal position, the particle is updated by Gaussian disturbance and local search strategy. The optimal particle will be selected for the next round of learning. This method not only enhances population diversity, but also increases optimizing precision. Simulations were performed on 12 benchmark functions under the same parameters. The results indicate that the optimizing performance of the proposed algorithm is superior to the other 5 recently provided FA methods. Local search strategy, as compared with random attraction model and Gaussian disturbance, can dramatically improve the optimizing performance.