Biogeography-based optimization with covariance matrix based migration

•Covariance matrix-based migration (CMM) is proposed.•CMM significantly enhances the rotational invariance of BBO.•A novel CMM-BBO approach is developed.•Numeric simulations show CMM-BBO effectively improves the performance of BBO.

Biogeography-based optimization (BBO) is a new evolutionary algorithm. The major problem of basic BBO is that its migration operator is rotationally variant, which leaves BBO performing poorly in non-separable problems. To overcome this drawback of BBO, in this paper, we propose the covariance matrix based migration (CMM) to relieve BBO’s dependence upon the coordinate system so that BBO’s rotational invariance is enhanced. By embedding the CMM into BBO, we put forward a new BBO approach, namely biogeography-based optimization with covariance matrix based migration, called CMM-BBO. Specifically, CMM-BBO algorithms are developed by the CMM operator being randomly combined with the original migration in various existing BBO variants. Numeric simulations on 37 benchmark functions show that our CMM-BBO approach effectively improves the performance of the existing BBO algorithms.

Graphical abstractFigure optionsDownload full-size imageDownload as PowerPoint slide