Junsuk Choe


Assistant Professor @ Department of Computer Science and Engineering,

(Affiliated) Assistant Professor @ Department of Artificial Intelligence, Sogang University,

(Affiliated) Chair @ Data Science and AI Major, Graduate School of Information & Technology,

Sogang University


Office: AS-913

Email: jschoe at sogang dot ac dot kr


Google Scholar / CV / Github / Calendar

I am an Assistant Professor in the Department of Computer Science and Engineering at Sogang University.  Before joining Sogang University in Fall 2021,  I spent two wonderful years as a Research Intern and then a Research Scientist at NAVER AI Lab, where I had the good fortune to work with Dongyoon Han, Sangdoo Yun, Seong Joon Oh, Sanghyuk Chun, Byeongho Heo, and Jung-Woo Ha.  I obtained my Ph.D. from the Yonsei University in 2020 under the supervision of Prof. Hyunjung Shim, and my B.S. also from the Yonsei University in 2013.  

For prospective students/members: We have two openings for the PhD or Integrated MS/PhD program. Please contact me via email if you are interested.

To Sogang University Undergraduate Students: We currently do not offer an undergraduate intern program. However, we supervise up to two undergraduate students each semester through the Research Project (연구프로젝트) course. If you are interested, please complete this form. Applications are reviewed in mid-June and mid-December, with notifications sent out by the end of those months regarding your enrollment status for the next semester.

News


10/2024: 1 paper accepted to NeurIPS 2024 Workshop on Adaptive Foundation Models

10/2024: Hyoseo is at Michigan State University as a visiting scholar, hosted by Prof. Sijia Liu (-4/2025)

9/2024: Granted the Global Humanities and Social Convergence Research Program (1.8B KRW)  as a Co-PI from the NRF Korea (PI: Prof. Dasaem Jeong)

8/2024: Dongjun received the NRF Fellowship for PhD students (20M KRW) from the NRF Korea

8/2024: Dongjun is at CMU as a visiting student, supported by the IITP AI Intensive Education Program (-2/2025)

7/2024: 1 paper accepted to Patter Recognition (IF: 5.1)

7/2024: Hyoseo started her research internship at NAVER AI Lab

4/2024: Granted the Outstanding Young Scientist Grant (900M KRW) as a PI from the NRF Korea (Co-PI: Prof. Buru Chang).

4/2024: 1 paper accepted to Pattern Recognition Letters (IF: 5.1)

12/2023: Received the Sogang College of Engineering Young Fellow Award

12/2023: 1 paper accepted to AAAI 2024

8/2023: 1 paper accepted to WACV 2024

7/2023: 1 paper accepted to ICCV 2023

7/2023: Junsuk, Minyoung, and Hyoseo served as visiting scholars at the Tübingen AI Center (Host: Prof. Seong Joon Oh and Prof. Zeynep Akata)

3/2023: 1 paper accepted to Pattern Recognition Letters (IF: 4.757)

10/2022: Received the Sogang College of Engineering Outstanding  Lecture Award. link
4/2022: 1 paper accepted to PAMI (IF:24.314)

2/2022: 1 paper accepted to CVPR 2022

9/2021: Joined the Dept. of CSE at Sogang University as an Assistant Professor

7/2021: 4 papers accepted to ICCV 2021

6/2021 : Recognized as an outstanding reviewer by CVPR 2021

3/2021: 1 paper accepted to PR (IF:7.196)

2/2021: 1 paper accepted to CVPR 2021

5/2020: 1 paper accepted to PAMI (IF: 17.861)

2/2020: 1 paper accepted to CVPR 2020

7/2019: 1 paper accepted to ICCV 2019 (oral presentation)

3/2019: 1 paper accepted to CVPR 2019 (oral presentation)

In 2018: 1 paper accepted to Information Sciences (IF: 5.524)

Research Interests

Data-centric Machine Learning highlights the significance of high-quality, well-labeled data over complex models. My work specifically targets enhancing model performance through better data preprocessing, augmentation, and labeling. By prioritizing data management, I aim to close the gap between theoretical models and practical applications.

Machine Learning with Minimal Supervision aims to efficiently use both labeled and unlabeled data to improve model performance while reducing the need for extensive manual labeling. Specifically, I am interested in weakly-supervised learning, which leverages weakly-labeled data, and continual learning, which enables models to learn and adapt over time without forgetting previous knowledge.

Trustworthy Machine Learning focuses on ensuring model reliability and transparency, which are crucial for developing robust and accountable AI systems. Specifically, I am working on machine unlearning to enable models to forget specific data points from their training set, thereby ensuring compliance with data privacy regulations and improving model adaptability. Additionally, I am focusing on uncertainty estimation to quantify the confidence of model predictions, providing insight into the reliability of the outputs and guiding decision-making processes

Lab Members

Dongjun Hwang (PhD student in CSE, 24.03-) 

Doyeol Baek (PhD student in CSE, 24.09-) 

Minyoung Lee (MS student in AI, 23.03-) 

Hyoseo Kim (Integrated MS-PhD student in CSE, 22.09-)

Yeji Park (Integrated MS-PhD student in CSE, 23.03-)

Yejin Kim (MS student in AI, 24.03-) 

Alumni

Doyeol Baek (MS student, 22.09-24.08) → PhD student at Sogang CSE

Sung-hwan Han (BS student, 23.01-24.06)

Dongjun Hwang (MS student, 22.03-24.02) → PhD student at Sogang CSE

Junyong Kang (BS student, 21.12-23.03) → MS student at KAIST AI

Teaching

Undergraduate

CSE4014: Introduction to Deep Learning (24F)

CSE4130: Introduction to Machine Learning (24S, 23S)

COR1010: Introduction to AI Programing (24S)

CSE3080: Data Structure (23S, 22S)

CSE4120: Fundamentals of Compiler Construction (22F, 21F)

CSE5321: Introduction to Natural Language Processing (23F)

CSE2003: Computer Programming I (22S)

CSE3015: Introduction to Digital Circuits (21F)

Graduate 

CSE6100: Deep Learning (24S)

CSE6517: Topics in Representation Learning (24F, 23F)

CSE6528: Bigdata Computing Capstone Design (24F, 23F, 22F)

GITA388: Introduction to Deep Learning (24F, 23F, 23S)

GITA388: Pattern Recognition (24S, 23S)

Submissions and Preprints

(† : Corresponding Author, *: Co-first Author)

Overcoming Domain Limitations in Open-vocabulary Segmentation [code]

Dongjun Hwang, Seong Joon Oh, Junsuk Choe† .

arXiv preprint: 2410.11536.

NegMerge: Consensual Weight Negation for Strong Machine Unlearning [code]

Hyoseo Kim, Dongyoon Han† , Junsuk Choe† .  

arXiv preprint: 2410.05583. (4-page summary published in NeurIPS 2024 Workshop on Adaptive Foundation Models)

Hyoseo's Internship Project at NAVER AI Lab

LMLT: Low-to-high Multi-Level Vision Transformer for Image Super-Resolution [code]

Jeongsoo Kim, Jongho Nang, Junsuk Choe† .

arXiv preprint: 2409.03516.

ConVis: Contrastive Decoding with Hallucination Visualization for Mitigating Hallucinations in Multimodal Large Language Models [code]

Yeji Park*, Deokyeong Lee*, Junsuk Choe† , Buru Chang† . 

arXiv preprint: 2408.13906.

Fog-Free Training for Foggy Scene Understanding

Minyoung Lee, Kyungwoo Song, Junsuk Choe† .

Submitted to Pattern Recognition Letters.

VHOIP: Video-based Human-Object Interaction recognition with CLIP Prior knowledge

Doyeol Baek, Junsuk Choe† .

Submitted to Pattern Recognition Letters (Under Revision).

Improved Semantic Segmentation by Fisheye Image Augmentation for Driving Scene

Hyeseong Lee, Sunmin Park, Miyoung Lee, Junsuk Choe† , Sungwon Jung† . 

Submitted to Pattern Recognition Letters.

Curriculum Learning with Class-label Composition for Weakly Supervised Semantic Segmentation

Dongjun Hwang, Hyoseo Kim, Doyeol Baek, Hyunbin Kim, Inhye Kye, Junsuk Choe† .

Submitted to Pattern Recognition Letters (Under Revision).

Improving ViT Interpretability with Patch-level Mask Prediction

Junyong Kang, Byeongho Heo, Junsuk Choe† .

Submitted to Pattern Recognition Letters (Under Revision).

Forecasting Effective Label Types for Weakly-supervised Semantic Segmentation

Minhyun Lee*, Kyungmin Kim*, Junsuk Choe, Hyunjung Shim† . 

Submitted to International Journal of Computer Vision.

Small Object Matters in Weakly Supervised Object Localization

Dongjun Hwang, Seong Joon Oh, Junsuk Choe† .

Submitted to Neurocomputing (Under Revision).

Selected Publications (2019-)

(† : Corresponding Author, *: Co-first Author)

Discovering an Inference Recipe for Weakly-Supervised Object Localization

Sanghuk Lee*, Cheolhyun Mun*, Youngjung Uh, Junsuk Choe, Hyeran Byun† . 

Patter Recognition, 2024.  (IF: 7.5) 

Weakly-supervised Incremental learning for Semantic segmentation with Class Hierarchy

Hyoseo Kim, Junsuk Choe† .

Pattern Recognition Letters, 2024.  (IF: 5.1) 

Weakly Supervised Semantic Segmentation for Driving Scenes [code]

Dongseob Kim*, Seungho Lee*, Junsuk Choe, Hyunjung Shim† .

AAAI Conference on Artificial Intelligence (AAAI), 2024.

Small Objects Matters in Weakly-supervised Semantic Segmentation

Cheolhyun Mun*, Sanghuk Lee*, Youngjung Uh, Junsuk Choe, Hyeran Byun† .

IEEE Winter Conference on Applications of Computer Vision (WACV) , 2024.

Neglected Free Lunch - Learning Image Classifiers Using Annotation Byproducts [code]

Dongyoon Han*, Junsuk Choe*, Seonghyeok Chun, John Chung, Minsuk Chang, Sangdoo Yun, Jean Song, Seong Joon Oh† .

International Conference on Computer Vision (ICCV), 2023.

Entropy Regularization for Weakly Supervised Object Localization

Dongjun Hwang, Jung-Woo Ha, Hyunjung Shim, Junsuk Choe† .

Pattern Recognition Letters, 2023.  (IF: 4.757) 

Evaluation for Weakly Supervised Object Localization: Protocol, Metrics, and Datasets [arxiv][code]

Junsuk Choe*, Seong Joon Oh*, Sanghyuk Chun, Seungho Lee, Zeynep Akata, Hyunjung Shim† . 

IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 2023.  (IF: 24.314) 

Weakly Supervised Semantic Segmentation using Out-of-Distribution Data [code]

Jungbeom Lee, Seong Joon Oh, Sangdoo Yun, Junsuk Choe, Eunji Kim, Sungroh Yoon† .

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 

Attention-based Dropout Layer for Weakly Supervised Single Object Localization and Semantic Segmentation [code]

Junsuk Choe*, Seungho Lee*, and Hyunjung Shim† . 

IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 2021.  (IF: 24.314)

Keep CALM and Improve Visual Feature Attribution [code]

Jae Myung Kim*, Junsuk Choe*, Zeynep Akata, Seong Joon Oh† . 

International Conference on Computer Vision (ICCV), 2021.

Normalization Matters in Weakly Supervised Object Localization [code]

Jeesoo Kim, Junsuk Choe, Sangdoo Yun, Nojun Kwak† .

International Conference on Computer Vision (ICCV), 2021.

Contrastive Attention Maps for Self-supervised Co-localization

Minsong Ki, Youngjung Uh, Junsuk Choe, Hyeran Byun† .

International Conference on Computer Vision (ICCV), 2021.

Rethinking Spatial Dimensions of Vision Transformers [code]

Byeongho Heo, Sangdoo Yun, Dongyoon Han, Sanghyuk Chun, Junsuk Choe, Seong Joon Oh.

International Conference on Computer Vision (ICCV), 2021.

Region-based Dropout with Attention Prior for Weakly Supervised Object Localization 

Junsuk Choe, Dongyoon Han, Sangdoo Yun, Jung-Woo Ha, Seong Joon Oh, Hyunjung Shim† . 

Pattern Recognition (PR), 2021.  (IF: 7.196) 

Re-labeling ImageNet: from Single to Multi-Labels, from Global to Localized Labels [code]

Sangdoo Yun, Seong Joon Oh, Byeongho Heo, Dongyoon Han, Junsuk Choe, Sanghyuk Chun.

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2021. 

Evaluating Weakly Supervised Object Localization Methods Right [code]

Junsuk Choe*, Seong Joon Oh*, Seungho Lee, Sanghyuk Chun, Zeynep Akata, Hyunjung Shim† . 

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2020. 

CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features [code]

Sangdoo Yun, Dongyoon Han, Seong Joon Oh, Sanghyuk Chun, Junsuk Choe, Youngjoon Yoo.

International Conference on Computer Vision (ICCV), 2019.  (Oral presentation)

Attention-based Dropout Layer for Weakly Supervised Object Localization [code]

Junsuk Choe, Hyunjung Shim† .  

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019.  (Oral presentation)