Seonguk Seo CV Scholar LinkedIn

I am a Ph.D. candidate in SNU Computer Vision Lab. under the supervision of Prof. Bohyung Han. My research interests mainly lie in reliable and trustworthy representation learning for model robustness, privacy, and transferability. I am also interested in interactive media art, ambient sound production, and brand experience design. I worked at Meta AI in 2022 and at Adobe Research in 2019 as a research scientist intern.


[Jul, 2024] Our machine unlearning paper, MUDA, is now available on arXiv.

[Jun, 2024] Our artwork <Notated Human> won the Red Dot Award: Design Concept 2024.

[May, 2024] Source code of FedRCL is now available.

[Mar, 2024] I will participate in CVPR 2024 Doctoral Consortium.

[Feb, 2024] Our paper about federated learning is accepted to CVPR 2024.

[Oct, 2023] I am co-organizing the 5th Large-scale Video Object Segmentation Challenge (Youtube-VOS) workshop in conjunction with ICCV 2023.

Education

Seoul National University Mar, 2018 - Current Ph.D., Computer Vision Lab., ECE
POSTECH Mar, 2010 - Feb, 2018 B.S., Valedictorian and Summa Cum Laude
TU Berlin Mar, 2017 - Aug, 2017 Visiting Student
Seoul Science High School Mar, 2007 - Feb, 2010

Work Experiences

Meta AI Research Scientist Intern May, 2022 - Sep, 2022 Cambridge, MA, USA
Adobe Research Research Intern Aug, 2019 - Oct, 2019 San Jose, CA, USA
VCNC Software Engineer Mar, 2014 - Feb, 2016 Seoul, Korea

Publications

2024
Revisiting Machine Unlearning with Dimensional Alignment
Seonguk Seo, Dongwan Kim, Bohyung Han
arXiv preprint
Relaxed Contrastive Learning for Federated Learning
Seonguk Seo*, Jinkyu Kim*, Geeho Kim*, Bohyung Han
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, WA, 2024
2023
Online Backfilling with No Regret for Large-Scale Image Retrieval
Seonguk Seo, Gokhan Uzunbas, Bohyung Han, Sara Cao, Joena Zhang, Taipeng Tian, Ser-Nam Lim
ICML Workshop on Localized Learning, Honolulu, HI, 2023
Group Robustness via Adaptive Class-Specific Scaling
Seonguk Seo, Bohyung Han
ICML Workshop on Spurious Correlations, Invariance and Stability, Honolulu, HI, 2023
2022
Unsupervised Learning of Debiased Representations with Pseudo-Attributes
Seonguk Seo, Joon-Young Lee, Bohyung Han
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, LA, 2022
InfoNeRF: Ray Entropy Minimization for Few-Shot Neural Volume Rendering
Mijeong Kim, Seonguk Seo, Bohyung Han
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, LA, 2022
Information-Theoretic Bias Reduction via Causal View of Spurious Correlation
Seonguk Seo, Joon-Young Lee, Bohyung Han
AAAI Conference on Artificial Intelligence (AAAI), Vancouver, Canada, 2022    
2020
URVOS: Unified Referring Video Object Segmentation Network with a Large-Scale Benchmark
Seonguk Seo, Joon-Young Lee, Bohyung Han
European Conference on Computer Vision (ECCV), Glasgow, UK, 2020
Learning to Optimize Domain Specific Normalization for Domain Generalization
Seonguk Seo, Yumin Suh, Dongwan Kim, Geeho Kim, Jongwoo Han, Bohyung Han
European Conference on Computer Vision (ECCV), Glasgow, UK, 2020
2019
Domain-Specific Batch Normalization for Unsupervised Domain Adaptation
Woong-Gi Chang*, Tackgeun You*, Seonguk Seo*, Suha Kwak, Bohyung Han
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, CA, 2019
Learning for Single-Shot Confidence Calibration in Deep Neural Networks through Stochastic Inferences
Seonguk Seo*, Paul Hongsuck Seo*, Bohyung Han
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, CA, 2019

Awards and Honors

CVPR Doctoral Consortium Jun, 2024
Red Dot Award: Design Concept Jun, 2024
KCCV Doctoral Colloquium Aug, 2023
AAAI Student Scholarship Feb, 2022
IPIU Excellence Paper Award Feb, 2022
Youlchon AI Star Scholarship Aug, 2020
Visual Domain Adaptation Challenge @ ICCV 2019 Nov, 2019
Samsung AI Forum 2019 Best Poster Award Nov, 2019
Samsung AI Forum 2018 Best Poster Award Sep, 2018

Academic Services

Organizer
The 5th Large-Scale Video Object Segmentation Challenge Workshop @ ICCV 2023
Oct, 2023
The 4th Large-Scale Video Object Segmentation Challenge Workshop @ CVPR 2022
Jun, 2022
The 3rd Large-Scale Video Object Segmentation Challenge Workshop @ CVPR 2021
Jun, 2021
Invited Talk
KCCV, Naver Labs, XAIC, IEIE
Reviewer
Journal: TPAMI, IJCV
Conference: ICLR, ICML, NeurIPS, CVPR, ICCV, ECCV, AAAI

Art Portfolios

<Notated Human> Interactive Media Art May / Jan, 2024 Funded by NCA and presented at the Seouliteum / Kepco art center, Seoul Winner of Red Dot Award: Design Concept 2024
<Code: Exposing Identity> Interactive Media Pavilion Art Oct, 2023 Funded by ARKO and presented at PLATFORM-L, Seoul
<Alain> Interactive Performance Art Jul, 2023 Funded by ARKO and performed at Theater QUAD, Seoul
<Artificial Shadows> Interactive Media Art Jul, 2023 - Aug, 2023 Funded by ARKO and presented at Gallery Sejul, Seoul
<Code: Inherent Identity> Interactive Media Pavilion Art Dec, 2022 Funded by ARKO and presented at Gallery Min, Seoul
<Vision Machine> Interactive Performance Art Nov, 2022 Funded by ARKO and filmed offline
<Holy Motors: The Vehicled Bodies> Interactive Performance Art Sep, 2022 - Oct, 2022 Funded by Hyundai ZER01NE and performed at S Factory, Seoul
<Disorient: Ringularity> Interactive Media Art Sep, 2022 - Oct, 2022 Funded by Soorim Cultural Fdn. and presented at Kim Hee-Su art center, Seoul
<Code: Registering Identity> Interactive Media Art Nov, 2021 Funded by ARKO and presented online
<Disorient: Ergo-sphere> Interactive Media Art Oct, 2021 Funded by KOCCA and presented at maru art center, Seoul
<Vender> Interactive Installation Art Dec, 2020 - Jan, 2021 Funded by and presented at art center nabi, Seoul