I. Introduction
Complex-Valued signals naturally arise in many applications like communication [1], signal processing [2]–[5], image processing [6]–[9], and radar applications [10]. Representing these signals and their nonlinear transformations in the complex domain is a natural way to preserve their physical characteristics. But, the analysis of complex-valued signals and their transformations in an efficient way pose many new challenges. Since the complex-valued neural networks (CVNNs) provide a convenient mechanism for the above, they have created renewed interest in researchers for developing new complex-valued neural architectures and their learning algorithms.