AI for Manufacturing

AI for Manufacturing

Optimization and ML play a key role in both the pre-silicon design and post-silicon control of electronic circuits. This end-use case  focuses on specialized algorithms in these areas, with hardware and algorithm domain experts working in close collaboration.

For pre-silicon design, the Institute explores data-driven techniques for automatic synthesis and optimization of analog mixed-signal (AMS) and digital circuits, mm-wave circuits with 3D electromagnetic (EM) structures, as well as heterogeneously integrated packaged components.

Beyond the design phase, online optimization and RL play a critical role in the “post-silicon” control of deployed massively scaled and reconfigurable electronics to maximize their efficiency and resiliency. Controlling active components (e.g., power amplifiers and front-end filters) as their operational environment changes can significantly extend their range of linearity and efficiency. 

Publications

Related work

  • Self-Supervised Policy Adaptation during Deployment,
    Nicklas Hansen, Yu Sun, Pieter Abbeel, Alexei A. Efros, Lerrel Pinto, Xiaolong Wang.
    In the proceedings of the 7th International Conference on Learning Representations (ICLR), Virtual, April 2021.
    arXiv 2007.04309

  • See the Future through the Void: Active Pre-Training with Successor Features,
    Hao Liu, Pieter Abbeel.
    In the proceedings of the International Conference on Machine Learning (ICML), Virtual, July 2021.

  • State Entropy Maximization with Random Encoders for Efficient Exploration,
    Younggyo Seo, Lili Chen, Jinwoo Shin, Honglak Lee, Pieter Abbeel, Kimin Lee.
    In the proceedings of the International Conference on Machine Learning (ICML), Virtual, July 2021.
    arXiv 2102.09430

  • Mutual Information-based State-Control for Intrinsically Motivated Reinforcement Learning,
    Rui Zhao, Yang Gao, Pieter Abbeel, Volker Tresp, Wei Xu. In the proceedings of the 7th International Conference on Learning Representations (ICLR), Virtual, April 2021. arXiv 2103.08107

  • coupling Representation Learning from Reinforcement Learning,
    Adam Stooke, Kimin Lee, Pieter Abbeel, Michael Laskin.
    In the proceedings of the International Conference on Machine Learning (ICML), Virtual, July 2021.
    arXiv 2009.08319

  • In-field performance optimization for mm-wave mixed- signal doherty power amplifiers: a bandit approach. S. Xu, F. Wang, H. Wang, and J. Romberg.  IEEE Trans. Circuits and Systems, to appear, 2020.

  • An artificial-intelligence (AI) assisted mm-wave doherty power amplifier with rapid mixed-mode in-field performance optimization. F. Wang, S. Xu, J. Romberg, and H. Wang. In Proc. IEEE IMC-5G, 2019.

  • A hardware realization of superresolution combining random coding and blurring. K. Beale, J. Chen, A. Giljjum, K. F. Kelly, and J. Romberg. IEEE Trans. Comput. Imaging, 5(3):366–380, 2019.

AI for Manufacturing Team

Borivoje Nikolic
Borivoje Nikolic
University of California, Berkeley
Leader
Hua Wang
Hua Wang
Georgia Institute of Technology
Co-leader
Pieter Abbeel
Pieter Abbeel
University of California, Berkeley
Jacob Abernathy
Jacob Abernathy
Georgia Institute of Technology
George Lan
George Lan
Georgia Institute of Technology
Siva Theja Maguluri
Siva Theja Maguluri
Georgia Institute of Technology
Arijit Raychowdhury
Arijit Raychowdhury
Georgia Institute of Technology
Justin Romberg
Justin Romberg
Deputy Director
Georgia Institute of Technology
Vladimir Stojnovic
Vladimir Stojnovic
University of California, Berkeley
Tuo Zhao
Tuo Zhao
Georgia Institute of Technology