Particle swarm optimisation with dynamic search strategies based on rank correlation
by Toshiki Nishio; Jun-ichi Kushida; Akira Hara; Tetsuyuki Takahama
International Journal of Computational Intelligence Studies (IJCISTUDIES), Vol. 6, No. 4, 2017

Abstract: The paper presents particle swarm optimisation with dynamic search strategies (DSS-PSO). In order to control search strategies, we introduce landscape modality estimation method using correlation coefficients between rankings of search points to PSO. This estimation method utilises relationship between fitness and distance to a reference points. By using estimated result, we stochastically classify whether the landscape modality is uni-modal or multi-modal and population is moving toward outside or not. Our proposal method can switch the strategies properly according to the landscape modality of an objective function. If the landscape modality is estimated as multi-modal, elitist learning strategy is performed. If the landscape modality is estimated as ridge, multi-dimensional mutation and mutation by differential vector are performed. Furthermore, inertia weight for velocity vector is controlled according to the estimated landscape modality. To confirm the search ability of the proposal method, we conducted experiments using standard benchmark functions. The experimental results show that the proposal method outperforms other PSO variants.

Online publication date: Mon, 29-Jan-2018

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Computational Intelligence Studies (IJCISTUDIES):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com