Machine learning algorithms and artificial intelligence (AI) have driven a revolution in many scientific and engineering fields. Now, they are slowly making their way into the field of electronic design automation (EDA). Bayesian learning (BL) has emerged as a promising approach to modeling and optimization for systems when the data is scarce and computational resources are strictly limited. Using BL, it is possible to build more accurate models by collecting data samples adaptively. This paper presents an overview of popular machine learning strategies that have been developed to automate different steps of the electronic design process. Particular attention is paid to Bayesian learning and how it is applied to the modeling and optimization of analog devices. In addition, this paper analyses some recent BL variants proposed to tackle specific machine learning issues and solve application-specific tasks. © 2012 IEEE.
