Using neural network models for efficient GaAs MESFET time domain nonlinear circuit analysis

Abstract

This paper describes the applications of multilayer perceptrons for GaAs device and integrated circuit modeling, that were proposed to improve circuit simulation programs like SPICE. The obtained results by the MESFET logic gate circuit simulations prove the validation of neural models applicability for the time domain nonlinear circuit transient and sensitivity analyses.