This chapter is divided into two parts. The first part discusses the different approaches to automate the process of sizing analog Integrated Circuits (ICs). These approaches in recent years have been evolving from more simple approaches like knowledge-based (that will not be discussed in this chapter), where automation tries to mimic manual designers by encoding the design flow in rules and equations, to more complex approaches using Optimizations or Machine Learning (ML) algorithms. This chapter is divided into two parts. The first part discusses the different approaches to automate the process of sizing analog ICs. These approaches in recent years have been evolving from more simple approaches like knowledge-based (that will not be discussed in this chapter), where automation tries to mimic manual designers by encoding the design flow in rules and equations, to more complex approaches using Optimizations or ML algorithms. The field of ML encompasses three primary domains. Firstly, there’s Unsupervised Learning, where when provided with an unlabeled dataset, train discovers inherent relationships and representations within. The goal of this field is to uncover hidden structures, or label unlabeled data. Following is the field of Reinforcement Learning (RL), involving the creation of an agent tasked with taking actions within an environment. Through these actions, the agent receives feedback in the form of rewards, guiding its ongoing interactions with the environment. Lastly, there’s Supervised Learning (SL) which involves training an algorithm with a labeled dataset by constructing a model capable of classification or regression. This trained model is then applied to new datasets, enabling informed predictions and analysis. It is also worth mentioning that the field of ML has some secondary domains, like Semi-supervised or Self-Supervised. The second part of this chapter will focus on the state of the art of one type of Self-Supervised Learning, the diffusion models. Its general implementations, and the different approaches of this methodology.