Career Profile

Yonghwi Kwon is currently a PhD student in School of Electrical Engineering, KAIST, Korea. Under supervision of Prof. Youngsoo Shin, his main research area is machine learning guided physical design, computational lithography, and low power design.

Education

PhD in School of Electrical Engineering

2019 - Present
Korea Advanced Institute of Science and Technology (KAIST)

MS & PhD integrated program

MS in School of Electrical Engineering

2018 - 2019
Korea Advanced Institute of Science and Technology (KAIST)

BS in School of Electrical Engineering

2014 - 2018
Korea Advanced Institute of Science and Technology (KAIST)

Projects

Dynamic voltage drop prediction with CNN (2019 - present)
Using U-net architecture, DvD of a layout is estimated with <15% error while it's >15x faster than conventional analysis.
Graph neural network guided leakage optimization (2019 - present)
GNN is adopted for faster Vth optimization on ECO stage for leakage power reduction without timing violation.
Gate-level clock gating for cyclic logic paths (2019 - present)
Gating function of cyclic logic paths are identified to use as enable signals of clock gating cells.
RNN based OPC acceleration (2018 - present)
To reflect causality of segment correction process in MB-OPC, RNN is applied for accurate mask bias prediction.
Synthetic test pattern generation using generative model (2018 - present)
To enhance the coverage of lithography model, synthetic patterns are generated using GANs.
Machine learning based SRAF printing prediction (2018 - 2019)
Project with SK Hynix R&D. ML-based SRAF printing prediction for full chip scale.
Early stage clock network power estimation (2017 - 2019)
Predicting clock network power with pre-CTS netlist and CTS parameters using ANN.

Publications

  • [C11] Test pattern clustering for accurate and efficient lithography modeling (submitted)
  • Gangmin Cho, Yonghwi Kwon, Pervaiz Kareem, and Youngsoo Shin
    IEEE/ACM Design Automation Conference (DAC), 2021
  • [C10] Dynamic IR drop prediction using image-to-image translation neural network (submitted)
  • Yonghwi Kwon, Daijoon Hyun, Giyoon Jung, and Youngsoo Shin
    IEEE International Symposium on Circuits and Systems (ISCAS), 2021
  • [C9] Test pattern extraction for lithography modeling under design rule revisions (accepted)
  • Gangmin Cho, Yonghwi Kwon, Pervaiz Kareem, Sungho Kim, and Youngsoo Shin
    SPIE Advanced Lithography, 2021
  • [C8] Fast prediction of process variation band through machine learning models (accepted)
  • Pervaiz Kareem, Yonghwi Kwon, Gangmin Cho, and Youngsoo Shin
    SPIE Advanced Lithography, 2021
  • [C7] Optimization of accurate resist kernels through convolutional neural network (accepted)
  • Yonghwi Kwon and Youngsoo Shin
    SPIE Advanced Lithography, 2021
  • [J2] Optical proximity correction using bidirectional recurrent neural network with attention mechanism (in revision)
  • Yonghwi Kwon, and Youngsoo Shin
    IEEE Transactions on Semiconductor Manufacturing (TSM)
  • [C6] Fast ECO leakage optimization using graph convolutional network (Best paper award candidate)
  • Wonjae Lee, Yonghwi Kwon, and Youngsoo Shin
    ACM Great Lakes Symposium on VLSI (GLSVLSI), Aug. 2020
  • [C5] Pre-layout clock tree estimation and optimization using artificial neural network
  • Sunwha Koh, Yonghwi Kwon, and Youngsoo Shin
    ACM/IEEE International Symposium on Low Power Electronics and Design (ISLPED), Sep. 2020
  • [J1] Layout pattern synthesis for lithography optimizations
  • Pervaiz Kareem, Yonghwi Kwon, and Youngsoo Shin
    IEEE Transactions on Semiconductor Manufacturing (TSM), vol. 33, May 2020
  • [C4] SRAF printing prediction using artificial neural network
  • Yonghwi Kwon, Jinho Yang, Sungho Kim, Cheolkyun Kim, and Youngsoo Shin
    SPIE Advanced Lithography, Feb. 2020
  • [C3] Clock gating synthesis of netlist with cyclic logic paths
  • Yonghwi Kwon, Inhak Han, and Youngsoo Shin
    IEEE/ACM International Conference on Computer-Aided Design (ICCAD), Nov. 2019
  • [C2] Optical proximity correction using bidirectional recurrent neural network (BRNN)
  • Yonghwi Kwon, Youngsoo Song, and Youngsoo Shin
    SPIE Advanced Lithography, Feb. 2019
  • [C1] Transient clock power estimation of pre-CTS netlist
  • Yonghwi Kwon, Jinwook Jung, Inhak Han, and Youngsoo Shin
    IEEE International Symposium on Circuits and Systems (ISCAS), May 2018

    Honors

    Best paper award nomination @ ACM Great Lakes Symposium on VLSI (GLSVLSI)
    Aug. 2020
    Invited speaker @ Optical Society Korea (OSK) Summer Meeting
    Jul. 2020
    Excellence award @ SK Hynix Open Idea Contest
    Nov. 2019, awarded with ~18k$ funding.
    Invited speaker @ Next Generation Lithography (NGL) conf.
    Aug. 2019, presentation about BRNN-OPC.
    Richard Newton Young Fellow @ 56th Design Automation Conf. (DAC)
    Jun. 2019

    Volunteer Acitivty

    Hanwha-KAIST Education Program for Gifted Students

    2019 - 2020
    Head mentor

    Program management for 80 students and 12 mentors.

    KAIST Science Outreach Program (KSOP)

    2015 - 2019
    Student mentor

    Bi-weekly teaching science and mathmatics to underpriviledged student group with talent on science. (Funded by Ministry of Science and ICT)

    Programming languages

    Python

    Tcl/Tk

    Unix shell scripting

    C

    BASIC

    MATLAB

    Verilog

    Tool proficiency

    TensorFlow

    PyTorch

    Synopsys Proteus & S-litho

    Synopsys ICC & DC & PT

    Mentor Graphics Calibre