Particle Swarm Extension to LOLIMOT

In this paper, we will present population based method for placement of center of radial basis function of a locally linear neuro-fuzzy (LLNF) network, which is trained by LOLIMOT algorithm. Originally, LOLIMOT algorithm incrementally divides the hyper-rectangles on input space into two axes orthogonal directions in half. However, this heuristic method would not be the best possible partitioning of the space. We present and evaluate a new particle swarm optimization (PSO) method for finding the best divisions of input space in the LOLIMOT algorithm