Machine-learning/AI-based Automated Layout Synthesis of Analog and Mixed-Signal Integrated Circuits

True automation of analog/mixed-signal (AMS) and RF integrated circuit layout design has long been a target in electronic design automation (EDA). Making good use of human designer heuristics in existing layouts that steer today’s tools is key to balancing efficiency and design space exploration, especially for analog circuit layouts. However, explicitly getting the constraints for every circuit from designers is the weak spot. In this project, various automated constraint learning and transfer based methods will be explored to solve this problem, learning how to lay out new circuits from existing expert-crafted layouts.