Stijn Ringeling's thesis reports ML-assisted acceleration of sigma-delta-modulator linearity evaluation

Stijn Ringeling's Eindhoven University of Technology thesis reports model-based methods for nonlinear discrete- and continuous-time sigma-delta modulators with a 5,000× evaluation acceleration in the behavioral-model studies. It also reports a machine-learning behavioral model and transfer learning to transistor-level data across supply variations using 20 circuit simulations, a dataset 50× smaller than the behavioral-model dataset. No company deployment is claimed.