Machine Learning Assisted Analog Circuit Design Automation

Abstract

Unlike digital integrated circuit design, analog integrated circuit design process generally lacks enough automation and relies heavily on the expertise and intuition of the analog designer. Attempts have been made in the past to automate several aspects of the analog design process, but these techniques never became mainstream due to the complexity of analog design. With the recent advancement of artificial intelligence (AI) and machine learning (ML) techniques, which promise to automate many complex tasks, there is a renewed interest in automating analog design process through AI / ML techniques. With such automation, cost and design time can be significantly reduced and new analog designers can be trained more efficiently. In this research, we develop an ML-based framework for automating analog design optimization at the schematic level.