A Transfer Learning Approach to Expedite Training of Artificial Neural Networks for Variability-Aware Signal Integrity Analysis of MWCNT Interconnects

Abstract

In this paper, an artificial neural network (ANN) trained using a novel transfer learning approach is presented for the variability-aware signal integrity analysis of on-chip multi-walled carbon nanotube (MWCNT) interconnects. In the proposed transfer learning approach, initially a secondary ANN is trained to emulate the signal integrity quantities of interest of an approximate equivalent single conductor (ESC) model of the MWCNT interconnects. Thereafter, the values of the weights and bias terms of this secondary ANN are used to expedite the training of the primary ANN that will emulate the signal integrity quantities of the more rigorous multiconductor circuit (MCC) model of the MWCNT interconnects.