Complex rival penalized learning for RBF neural network used in communication channel equalization

A complex rival penalized competitive learning is proposed for radial basis function neural network centers initialization. Performances are directly related to the clustering centers estimations. The network has complex centers and connection weights, but the nonlinearity of its hidden nodes remains a real-valued function. The radial basis function network is able to generate complicated nonlinear decision regions or to approximate an arbitrary nonlinear function in complex multidimensional space. For this aim the complex-valued radial basis neural network is proposed for digital communications channel equalization.