profile picture

Eunji Kwon

Assistant Professor at Kookmin University

Education

Ph.D. in Electrical Engineering

POSTECH
2021 - 2024

M.Sc. in Electrical Engineering

POSTECH
2019 - 2021

B.Sc. in Mechanical Engineering (Summa Cum Laude)

UNIST
2013 - 2017

Languages

  • Korean (Native)
  • English (Professional)

Interests

  • Deep Learning Hardware Accelerator
  • Deep Learning SW/HW Co-optimization
  • Power Management on Mobile Device Systems

About Theme

Assistant Professor at Kookmin University

Eunji Kwon received the B.S. degree in mechanical engineering in 2017 from Ulsan National Institute of Science and Technology (UNIST), Ulsan, South Korea. She received M.S. degree in electrical engineering in 2021 and received Ph.D. degree in electrical engineering in 2024 from Pohang University of Science and Technology (POSTECH), Pohang, South Korea. She is currently an assistant professor of artificial intelligence at Kookmin University. Her current research interest is energy-efficient AI computing.

Experiences

Graduate Student

Feb. 2021 - Aug. 2024
CSDL Laboratory in POSTECH, South Korea, Prof. Seokhyeong Kang

Visiting Graduate Student

Aug. 2023 - Feb. 2024
SEE Laboratory in University of California San Diego (UCSD), California, Prof. Tajana Losing

Undergraduate Student Research Intern

Jan. 2017 - Feb. 2017
ACRC Laboratory in WPI, Boston, Pdoc. Eunkyung

Projects

Project Lists.

Deep Learning and Development of Voice Interactive Smart Home Control System Based on Internet of Home Things (IoHT) - POSCO Future IT Convergence Research Institute, Dec.2017 - Dec.2020
Power/thermal analysis and management for MPSoC mobile device systems - Samsung Electronics, July.2014 - June.2021
Autonomous Accuracy Scaling for Ultra-low-power Intelligent Mobile Systems - Samsung Science & Technology Foundation, Aug.2017 - Jul.2020
Speaker Recognition during Business Meeting - POSCO ICT, Mar.2020 - Dec.2020
FPGA-based Object Detection Model for Unmanned Submersible - LIG Nex 1, June.2021 - Nov.2022
Low-power Vision Transformer Accelerator - Research Project Manager, National Research Foundation of Korea (NRF), June.2022 - May.2024

Publications

Eunji Kwon’s Publications (International Journals and Conferences)

RL-PTQ : RL-based Mixed Precision Quantization for Hybrid Vision Transformers
E. Kwon, M. Zhou, W. Xu, T. Losing and S. Kang
ACM/IEEE Design Automation Conference (DAC), 2024
Mobile Transformer Accelerator Exploiting Various Line Sparsity and Tile-based Dynamic Quantization
E. Kwon, J. Yoon and S. Kang
IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (TCAD), 2023
Mobile Accelerator Exploiting Sparsity of Multi-Heads, Lines, and Blocks in Transformers in Computer Vision
E. Kwon, H. Song, J. Park and S. Kang
ACM/IEEE Design Automation and Test in Europe (DATE), 2023
Reinforcement Learning-based Power Management Policy for Mobile Device Systems
E. Kwon, S. Han, Y. Park, J. Yoon and S. Kang
IEEE Transactions on Circuits and Systems, Regular Papers (TCAS-I), 2021
Late Breaking Results : Reinforcement Learning-based Power Management Policy for Mobile Device Systems
E. Kwon, S. Han, Y. Park, Y. H. Kim and S. Kang
ACM/IEEE Design Automation Conference (DAC), 2020
ViT-ToGo : Vision Transformer Accelerator with Grouped Token Pruning
S. Lee, K. Cho, E. Kwon, S. Park, S. Kim and S. Kang
Design Automation and Test in Europe (DATE), 2024
Outlier-aware Time-multiplexing MAC for Higher Energy-efficiency on CNNs
E. Kwon, Y. Kang and S. Kang
International SoC Design Conference (ISOCC), 2019
FPGA-Based Accelerator for Rank-Enhanced and Highly-Pruned Block-Circulant Neural Networks
H. Song, J. Yoon, D. Kim, E. Kwon and S. Kang
Design Automation and Test in Europe (DATE), 2023
Adaptive FSP : Adaptive Architecture Search with Filter Shape Pruning
A. Kim, S. Lee, E. Kwon and S. Kang
16th Asian Conference on Computer Vision (ACCV), 2022
Approach to Improve the Performance using Bit-level Sparsity in Neural Networks
Y. Kang, E. Kwon, S. Lee, Y. Byun, Y. Lee and S. Kang
ACM/IEEE Design Automation and Test in Europe (DATE), 2021
MDARTS : Multi-objective Differentiable Neural Architecture Search
S. Kim, H. Kwon, E. Kwon, Y. Choi, T. Oh and S. Kang
ACM/IEEE Design Automation and Test in Europe (DATE), 2021
GRLC : Grid-based Run-length Compression for Energy-efficient CNN Accelerator
Y. Park, Y. Kang, S. Kim, E. Kwon and S. Kang
ACM/IEEE International Symposium on Low Power Electronics and Design (ISLPED), 2020
Analysis and Solution of CNN Accuracy Reduction over Channel Loop Tiling
Y. Kang, Y. Park, S. Kim, E. Kwon, T. Lim, S. Oh, M. Woo, and S. Kang
ACM/IEEE Design Automation and Test in Europe (DATE), 2020

Skills & Proficiency

Python/C

Pytorch

Verilog

CUDA

Xilinx Vivado

Xilinx Vivado HLS

Xilinx Petalinux

Cadence NC-Verilog/Simvision

Synopsys Design Compiler