Abstract:The work presented in this dissertation belongs to the scientific area of electronic design automation and addresses the automatic sizing of analog integrated circuits. Particularly, this work explores an innovative approach to automatic circuit sizing using deep learning and artificial neural networks to learn patterns from previously optimized design solutions. In opposition to classical optimization-based sizing strategies, where computational intelligent techniques are used to iterate over the map from devices’ sizes to circuits’ performances provided by design equations or circuit simulations, artificial neural networks are shown to be capable of solving analog integrated circuit sizing as a direct map from specifications to the devices’ sizes. Two separate artificial neural network architectures are proposed: a Regression-only model and a Classification and Regression model. The goal of the Regression-only model is to learn design patterns from the studied circuits, using circuit’s performances as input features and devices’ sizes as target outputs. This model can size a circuit given its specifications for a single topology. The Classification and Regression model has the same capabilities of the previous model, but it can also select the most appropriate circuit topology and its respective sizing given the target specification. The proposed methodology was implemented and tested on two analog circuit topologies. The achieved results show that the trained artificial neural networks were able to extend the circuit performance boundaries outside the train/ validation set, showing that, more than a mapping from the training data, the model is actually capable of learning reusable design patterns. Keywords—Analog Integrated Circuit Design, Electronic Design Automation, Deep Learning, Artificial Neural Networks

An Efficient Analog Circuit Sizing Method Based on Machine Learning Assisted Global Optimization

Machine Learning Driven Global Optimisation Framework for Analog Circuit Design

Fast Surrogate-Assisted Constrained Multiobjective Optimization for Analog Circuit Sizing Via Self-Adaptive Incremental Learning

A Review of Machine Learning Techniques in Analog Integrated Circuit Design Automation

DC-Model: A New Method for Assisting the Analog Circuit Optimization

Procedural- and Reinforcement-Learning-Based Automation Methods for Analog Integrated Circuit Sizing in the Electrical Design Space

Circuit Optimization over Multiple Process Corners for Analog Electronic Design Automation.

Closing the Gap between Electrical and Physical Design Steps with an Analog IC Placement Optimizer Enhanced with Machine-Learning-Based Post-Layout Performance Regressors

Post-Layout Simulation Driven Analog Circuit Sizing

Design Optimization of MOS Operational Amplifiers Using Finite Difference Sensitivity

Analog Circuit Optimization System Based on Hybrid Evolutionary Algorithms

Analog Circuit Sizing Using Machine Learning Based Transistor Circuit Model

Analog/Mixed-Signal Circuit Synthesis Enabled by the Advancements of Circuit Architectures and Machine Learning Algorithms

Using Artificial Neural Networks to Size Analog Integrated Circuits

Graph Theory Based Machine Learning for Analog Circuit Design

Smart-MSP: A Self-Adaptive Multiple Starting Point Optimization Approach for Analog Circuit Synthesis

Analog Circuit Design Optimization Based on Evolutionary Algorithms

DNN-Opt: An RL Inspired Optimization for Analog Circuit Sizing using Deep Neural Networks

Topology Generic DC-Model for Accelerating Analog Circuit Optimization

High-Dimensional Analog Circuit Sizing via Bayesian Optimization in the Variational Autoencoder Enhanced Latent Space

Transfer learning with bayesian optimization-aided sampling for efficient AMS circuit modeling