AI-Powered Revolution: Redefining Analog Circuit Design and Optimization for Tomorrow

作     者：Kong, Zhi Hui James, Anusha Ata, Sezin Kircali Aung, Khin Mi Mi Foo, Chuan Sheng James, Ashish Yeo, Kiat Seng 

作者机构：Singapore Univ Technol & Design SUTD Singapore 487372 Singapore ASTAR Inst Infocomm Res I2R Singapore 138632 Singapore ASTAR Ctr Frontier AI Res CFAR Singapore 138632 Singapore Tianjin Univ TJU Sch Microelect Tianjin 300072 Peoples R China Singapore Univ Technol & Design SUTD Engn Prod Dev Singapore 300072 Singapore 

出 版 物：《IEEE NANOTECHNOLOGY MAGAZINE》 (IEEE Nanatechnol. Mag.)

年 卷 期：2025年第19卷第2期

页      面：38-47页

核心收录：

学科分类：0808[工学-电气工程] 08[工学] 0805[工学-材料科学与工程（可授工学、理学学位）] 

基　　金：Singapore National Research Foundation [NRF-CRP20-2017-0006] Agency for Science, Technology and Research (A*STAR) under its AME Programmatic Funds [A19E3b0099, A20H6b0151] 

主　　题：Artificial intelligence Analog circuits Circuit synthesis Data models Integrated circuit modeling Solid modeling Optimization Computational modeling Accuracy Supervised learning Design methodology Active learning analog circuit design artificial intelligence bayesian optimization data-centric design automation initialization operational amplifier regression 

摘      要：The infusion of Artificial Intelligence (AI) within the intricate realm of analog circuits presents a transformative opportunity. By seamlessly blending the precision of AI algorithms with the nuanced operation of analog components, the landscape of analog circuit design has undergone a revolutionary transformation, enabling dynamic adaptation, fine optimization, and intelligent assimilation of insights gleaned from past data. Nevertheless, analog circuits face stringent functional and technological constraints, leading to a scarcity of data for modelling, and additional data acquisition entails costly and time-consuming simulations. This article introduces novel methodologies designed to enhance simulation efficiency and reduce associated costs, thereby enabling efficient and effective AI-driven analog circuit design. By leveraging data-driven AI approaches, the focus is on exploration of promising and feasible circuit design regions, improving AI model accuracy, and substantially mitigating the reliance on extensive simulations and significant manual effort. The results demonstrate a marked advancement in analog circuit design, showcasing how data-centric AI approaches can refine the design process, making it more efficient and cost-effective. This work is poised to set the stage for future developments where analog circuit design can be conducted with greater precision and efficiency.