Robustness Validation of a mm-Wave Model based on GRU Neural Networks for a GaN Power HEMT

Recently a GaN HEMT small-signal model based on gated recurrent unit (GRU) neural networks has been proposed and it has been proved that a very good accuracy can be achieved by using the developed modeling approach. However, as the GRU networks are recurrent networks, meaning that the previous states contribute to the current output, the aim of the work is to test and validate the behavior of the GaN HEMT model depending on the input order and history in order to derive the corresponding outcomes and to gain a better understanding of the inherent features of the developed modeling strategy.