Complex Varying-Parameter Zhang Neural Networks for Computing Core and Core-EP Inverse

Abstract

An improved complex varying-parameter Zhang neural network (CVPZNN) for computing outer inverses is established in this paper. As a consequence, three types of complex Zhang functions (ZFs) which are used for computing the time-varying core-EP inverse and core inverse are given. The convergence rate of the proposed complex varying-parameter Zhang neural networks (CVPZNNs) is accelerated. The super-exponential performance of the proposed CVPZNNs with linear activation is proved. Also, the upper bounds of a finite time convergence which correspond to the proposed CVPZNN with underlying Li and tunable activation functions are estimated. The simulation results, which relate the CVPZNNs with different activation functions, are presented.