AI-EDA: Toward a Holistic Approach to AI-Powered EDA

AI or ML has been extensively studied in CAD and EDA. The number of papers under this topic, published in major conference proceedings and journals between 2016 and 2020, has almost doubled [1] accounting for 11% of all regular papers in 2020. This could be attributed to a number of reasons: development of a series of new powerful ML algorithms, to better address increasing scalability issue of integrated circuits and their design, and to explore a possibility of switch from traditional heuristic-driven to new data-driven EDA.