Multiphysics-Informed Machine Learning for AI-Driven RFIC Design

Prevailing Machine-Learning (ML)-based approaches leverage existing solvers and simulators to generate a rich set of solutions, which are then used to train a neural network to predict outcomes for unseen cases. This approach can be computationally costly and it does not guarantee accuracy for unseen RFIC design. In this talk, Prof. Jiao will present recent advances in multiphysics-informed data-free ML, where machine intelligence is fused with domain expertise for significantly accelerated modeling, analysis, and optimization for RFIC design.