Research Interests

• Inverse Problems and Computational Imaging

- Image restoration for Natural Images (Denoising, Super-Resolution, etc.)

- Medical Image Reconstructions (CT, Ultrasound, Microscopy)

• Multi-Modal Learning for AI Application

- Enhancing Multi-modal learning with Optimal Transport

- Large Multimodal model for Health Care AI

• Deep Generative AI

- Developing efficient generative model with theoretical understandings

- Accelerating Diffusion models

Education

• Ph. D., Kim Jae Chul Graduate School of AI  (Sep. 2020 - Aug. 2024)

  • Korea Advanced Institute of Science and Technology (KAIST), South Korea

    • • Thesis Title :   "Score Matching Based Self-Supervised Image Denoising and Parameter Estimation"

  • Advisor: Prof. Jong Chul Ye

• M. S., Electrical Engineering (Aug. 2018 - Aug. 2020)

  • Ulsan National Institute of Science and Technology (UNIST), South Korea

  • Thesis Title :   "Unsupervised Training of Denoisers for Low-Dose CT Reconstruction Without Full-Dose Ground Truth"

  • Advisor: Prof.  Se Young Chun, (He is currently in Seoul National University, SNU )

• B. S., Electrical Engineering (Feb. 2012 - Aug. 2018)

  • • • Ulsan National Institute of Science and Technology (UNIST), South Korea, Summa Cum laude

Work Experiences

• Sep. 2024 - Current AI Research Scientist, Samsung Reserach

UNICORN: Ultrasound Nakagami Imaging via Score Matching and Adaptation for Assessing Hepatic Steatosis

Kwanyoung Kim*, Jaa-Yeon Lee*,  Jong Chul Ye, 

under Medical Image Analysis Review (Co-first author)

• (KoSAIM), 2024, (Best Oral Presentation Award) [paper]

End-to-End Breast Cancer Radiotherapy Planning via LMMs with Consistency Embedding

Kwanyoung Kim*, Yujin Oh*,  Sangjoon Park*,  Hwa Kyung Byun, Jin Sung Kim, Yong Bae Kim, Jong Chul Ye

under Medical Image Analysis Review (Co-first author) [paper]

OTSeg: Multi-prompt Sinkhorn Attention for Zero-Shot Semantic Segmentation

Kwanyoung Kim*, Yujin Oh*, Jong Chul Ye, *equal contribution

ECCV'2024 (Co-first author)  [paper] [code]

Noise2One: One-Shot Image Denoising wtih Local Implicit Learning

Kwanyoung Kim,  Jong Chul Ye, 

ICASSP'2024 (Oral) (First author) [paper] [code]

Unpaired Image-to-Image Translation via Neural Schrödinger Bridge 

BeumSu Kim*,  Gihyun Kwon*, Kwanyoung Kim, Jong Chul Ye, *equal contribution

ICLR'2024 (Co-author) [paper] [code]

Noise Distribution Adaptive Self-Supervised Image Denoising using Tweedie Distribution and Score Matching

Kwanyoung Kim, Taesung Kwon, Jong-Chul Ye 

CVPR'2022 (First author) [paper] [code]

Noise2Score: Tweedie's Approach to Self-Supervised Image Denoising without Clean Images

Kwanyoung Kim, Jong Chul Ye

NeurIPS'2021 (First author) [paper] [code]

Task-Aware Variational Adversarial Active Learning 

Kwanyoung Kim, Dongwon Park, Kwang In Kim, Se Young Chun 

CVPR'2021 (First author) [paper] [code]

Unsupervised Training of Denoisers for Low-Dose CT Reconstruction Without Full-Dose Ground Truth 

Kwanyoung Kim, Shakarim Soltanayev, Se Young Chun 

IEEE Journal of Selected Topics in Signal Processing  (IF: 6.688), 2020, (First author) [paper]