An Efficient Algorithm for Complex-Valued Neural Networks Through Training Input Weights

Abstract

Complex-valued neural network is a type of neural networks, which is extended from real number domain to complex number domain. Fully complex extreme learning machine (CELM) is an efficient algorithm, which owes faster convergence than the common complex backpropagation (CBP) neural networks. However, it needs more hidden neurons to reach competitive performance. Recently, an efficient learning algorithm is proposed for the single-hidden layer feed-forward neural network which is called the upper-layer-solution-aware algorithm (USA). Motivated by USA, an efficient algorithm for complex-valued neural networks through training input weights (GGICNN) has been proposed to train the split complex-valued neural networks in this paper. Compared with CELM and CBP, an illustrated experiment has been done in detail which observes the better generalization ability and more compact architecture for the proposed algorithm.