Research Topics

Topic 1: Deep Neural Network Design Automation

In this project we will explore 1) DNN compression / acceleration and 2) DNN design space exploration.

approx-dnn 

Selected recent publications:

  • [C132] Qi Sun, Chen Bai, Tinghuan Chen, Hao Geng, Xinyun Zhang, Yang Bai, Bei Yu, “Fast and Efficient DNN Deployment via Deep Gaussian Transfer Learning”, IEEE International Conference on Computer Vision (ICCV), Oct. 11–17, 2021. (paper)

  • [C129] Yang Bai, Xufeng Yao, Qi Sun, Bei Yu, “AutoGTCO: Graph and Tensor Co-Optimize for Image Recognition with Transformers on GPU”, IEEE/ACM International Conference on Computer-Aided Design (ICCAD), Nov. 1–4, 2021. (paper)

  • [C114] Qi Sun, Tinghuan Chen, Siting Liu, Jin Miao, Jianli Chen, Hao Yu, Bei Yu, “Correlated Multi-objective Multi-fidelity Optimization for HLS Directives Design”, IEEE/ACM Proceedings Design, Automation and Test in Europe (DATE), Feb. 01–05, 2021. (paper) (slides) (Best Paper Award Nomination)



Topic 2: Deep Mask Learning

hotspots 

Selected recent publications:

  • [C127] Ran Chen, Shoubo Hu, Zhitang Chen, Shengyu Zhu, Bei Yu, Pengyun Li, Cheng Chen, Yu Huang, Jianye Hao, “A Unified Framework for Layout Pattern Analysis with Deep Causal Estimation”, IEEE/ACM International Conference on Computer-Aided Design (ICCAD), Nov. 1–4, 2021. (paper)

  • [C126] Hao Geng, Fan Yang, Xuan Zeng, Bei Yu, “When Wafer Failure Pattern Classification Meets Few-shot Learning and Self-Supervised Learning”, IEEE/ACM International Conference on Computer-Aided Design (ICCAD), Nov. 1–4, 2021. (paper)

  • [C125] Binwu Zhu, Ran Chen, Xinyun Zhang, Fan Yang, Xuan Zeng, Bei Yu, Martin D.F. Wong, “Hotspot Detection via Multi-task Learning and Transformer Encoder”, IEEE/ACM International Conference on Computer-Aided Design (ICCAD), Nov. 1–4, 2021. (paper)

  • [C124] Guojin Chen, Ziyang Yu, Hongduo Liu, Yuzhe Ma, Bei Yu, “DevelSet: Deep Neural Level Set for Instant Mask optimization”, IEEE/ACM International Conference on Computer-Aided Design (ICCAD), Nov. 1–4, 2021. (paper)

  • [C104] Guojin Chen, Wanli Chen, Yuzhe Ma, Haoyu Yang, Bei Yu, “DAMO: Deep Agile Mask Optimization for Full Chip Scale”, IEEE/ACM International Conference on Computer-Aided Design (ICCAD), Nov. 2–5, 2020. (paper) (slides) (whova)

  • [C57] Hang Zhang, Fengyuan Zhu, Haocheng Li, Evangeline F. Y. Young, Bei Yu, “Bilinear Lithography Hotspot Detection”, ACM International Symposium on Physical Design (ISPD), pp. 7–14, Portland, OR, Mar. 19–22, 2017. (paper) (Best Paper Award)



Topic 3: Learning in Design Automation

Machine learning is a powerful computer science technique which can derive knowledge from big data, and provides prediction and matching. Since nanometer VLSI CAD problems have extremely high complexity and gigantic data, there has been a surge recently in applying and adapting machine learning techniques in VLSI CAD.

hotspots 

Selected recent publications:

  • [C122] Chen Bai, Qi Sun, Jianwang Zhai, Yuzhe Ma, Bei Yu, Martin D.F. Wong, “BOOM-Explorer: RISC-V BOOM Microarchitecture Design Space Exploration Framework”, IEEE/ACM International Conference on Computer-Aided Design (ICCAD), Nov. 1–4, 2021. (paper)

  • [C121] Jianwang Zhai, Chen Bai, Binwu Zhu, Yici Cai, Qiang Zhou, Bei Yu, “McPAT-Calib: A Microarchitecture Power Modeling Framework for Modern CPUs”, IEEE/ACM International Conference on Computer-Aided Design (ICCAD), Nov. 1–4, 2021. (paper)



Topic 4: Hardware Friendly Computer Vision

hardware-cv 

Selected recent publications:

  • [C130] Jiequan Cui, Zhisheng Zhong, Shu Liu, Bei Yu, Jiaya Jia, “Parametric Contrastive Learning”, IEEE International Conference on Computer Vision (ICCV), Oct. 11–17, 2021. (paper)

  • [C128] Wenqian Zhao, Qi Sun, Yang Bai, Haisheng Zheng, Wenbo Li, Bei Yu, Martin D.F. Wong, “A High-Performance Accelerator for Super-Resolution Processing on Embedded GPU”, IEEE/ACM International Conference on Computer-Aided Design (ICCAD), Nov. 1–4, 2021. (paper)



Topic 5: Combinatorial Algorithms in Design Automation

Many classical VLSI CAD problems can be extracted and formulated into challenging combinatorial optimization problems. We are heavily working to improve the state-of-the-art.

hotspots 

Selected recent publications are listed as follows: