Efficient parameters selection for artificial intelligence models of nanoscale MOSFETs

In this paper, the effect of network type in modeling I-V characteristic of MOS transistors was studied. Neural networks training data are generated in Hspice environment for MOSFET BSIM3 with TSMC-0.18 technology. Training was performed in MATLAB environment while testing was done in Hspice as well. Also in this work, feature selection using UTA method is utilized for determining consistency of BSIM3 parameters in MOSFET drain current estimation.