Group based Swarm evolution algorithm (GSEA) driven mental task classifier

Brain computer interface (BCI) system is based on non-invasive electroencephalographic signals. It is an augmented communication channel for disabled persons which translates neural activity from the brain into control signals. In this paper, a novel group based Swarm evolution algorithm (GSEA) driven neural classifier has been proposed to classify mental tasks. These mental tasks consists of left and right hand movement imagination and thinking of word generation. GSEA is a hybrid of swarm intelligence based computational methods with differential evolution based techniques. Hybridization of both algorithms ensures that pitfalls of both are overcome thus enhancing results. A concept of grouping has also been introduced to increase both exploration and exploitation performance of the algorithm. GSEA has been tested on publicly available BCI Competition 3 dataset 5. Experimental results shows that the proposed algorithm exhibits better result than individual swarm or differential search based algorithms and many other algorithms.