Machine Learning Techniques for VLSI Circuit Design: A Review

Analog, digital, mixed signal and RF circuits are popular in medical, healthcare, transportation, consumer electronics, IT, mobile devices, and home security. High production cost and extended time-to-market are the major issues for the development of these chips. Traditional manual and heuristic methods for designing analog ICs are time-consuming. Even with high-end CAD tools, hand computations are required optimise speed, area, and power. With the advanced technology nodes and complex design, the chip integration and performance are the major challenges. Several recent studies have examined ways to integrate machine learning (ML) and Artificial Intelligence (AI) with analog IC design tools and processes. The semiconductor development cycle and EDA offer ML/AI solution development and integration opportunities. Intelligent learning algorithms may save design time, increase circuit performance, and eliminate chip manufacturing design errors. This article reviews the literature on ML methodologies used in VLSI circuit design. ML/AI techniques have been used for design and performance optimisation, analog behaviour modelling, autonomous synthesis, behaviour analysis and verification, layout creation and process variation compensation. This article provides a rigorous overview of the field's current state and future growth potential.