Causal Inference for AMS Design (U. of Florida)

A new technical paper titled "Causal AI For AMS Circuit Design: Interpretable Parameter Effects Analysis" by the University of Florida introduces a causal-inference framework for Analog-mixed-signal (AMS) circuit design. The framework utilizes a directed-acyclic graph (DAG) from SPICE simulation data to quantify parameter impact through Average Treatment Effect (ATE) estimation. This approach provides human-interpretable rankings of design knobs and 'what-if' predictions, enhancing designers' understanding of trade-offs in sizing and topology. The study evaluates the framework on three operational-amplifier families and demonstrates higher accuracy and explainability compared to a baseline neural network regressor.