Affiliations: [a]
Department of Computer and Information Science, Zunyi Normal College, Zunyi, China
| [b]
Guangxi High School Key Laboratory of Complex System and Computational Intelligence, Nanning, China
| [c]
Institute for Computational Science and Faculty of Civil Engineering, Ton Duc Thang University, Hochiminh City, Vietnam | [d]
Faculty of Information Technology, Industrial University of Hochiminh city, Hochiminh, Vietnam
Correspondence:
[*]
Corresponding author. Aijia Ouyang, Department of Computer and Information Science, Zunyi Normal College, Zunyi 563002, China. Tel./Fax: +86851 28927172; E-mail: [email protected].
Abstract: Considering the problems of slow convergence and easily getting into local optimum of invasive weed optimization (IWO) algorithm in finding the optimal solution to large scale global optimization (LSGO) problems, we have proposed an improved IWO (IIWO) algorithm on the basis of the basic IWO algorithm. Concrete adjustments include setting the newborn weed seeds per plant to a fixed number of parameters, changing the initial step and final step to adaptive step, and re-initializing the solution which exceeds the limit value. Meanwhile, through applying the IIWO algorithm to the GPU platform, a parallel IIWO (PIIWO) based on GPU is obtained. The algorithm not only improves the convergence rate, but also strikes a balance between the global and local search capabilities. The simulation results of solving on the LSGO problems (CEC’ 2010 high-dimensional functions), have shown that, compared with other algorithms, our designed IIWO can yield better performance, faster convergence speed and higher accuracy; whilst the PIIWO has fewer iterations, higher computing accuracy and significant speedup than the serial algorithm IIWO.
Keywords: Adaptive step, fixed population, invasive weed optimization, GPU, large scale global optimization, speedup
