Experience

Assistant Professor
Sejong University
Department of Software
Mar 2025 - Present
ML Research Scientist
Twelve Labs.
Video Foundation Model, Evaluation
Mar 2024 - Mar 2025
AI Research Engineer
Upstage
Document Understanding, Synthetic Dataset
Mar 2023 - Mar 2024
Visiting Researcher
Clova AI Research (CLAIR), Naver Corp.
Image-to-Image translation
Sep 2019 - Feb 2020
Developer
SIT, Yonsei University
Real-time Remote Lecture Room System
Sep 2018 - Feb 2019

Education

Ph.D. in Integrated Technology
Yonsei University
Advisor: Prof. Hyunjung Shim
Thesis: Data-efficient Learning for Generative Adversarial Networks
Mar 2018 - Feb 2023
B.S. in EE and CS (Double Major)
Yonsei University
Mar 2014 - Feb 2018

Skills

Python (PyTorch) C / C++ Java SQL OpenCV / CUDA

Contact

Feel free to reach out to me via email.

Teaching

Sejong University
Instructor
• Generative AI
• Probability and Statistics
• Computer Architecture
• Creative Studies with Self-Directed Learning
Yonsei University
Teaching Assistant
• Computational Thinking and SW Programming
• Tensorflow for computer vision

Activities

Academic Services

Judge
• Academic festival at AI
• Hackerthon at Dept. of SW

External Activities

Reviewer
CVPR, ICCV, ECCV, NeurIPS, AAAI, IJCAI, WACV, BMVC, 3DV, ICCV-W, ICML-W, NeurIPS-W, CVPR-W, TPAMI, IJCV, TCSVT, etc.
Invited Talks
• Samsung Electronics DS DIT
• OKESTRO A.I. Lab/Data Science

Recent Preprints
(* : equal contribution. / † : corresponding author.)

From Solipsistic Simulators to Shared Reality: Multi-User World Models and the Consistency Trilemma
Kyungjune Baek*†, Jiho Jang*, Nojun Kwak
Preprint 2026, Under Review
Gradient-Driven Gaussian Initialization: A Scalable Framework for 2D Image Representation via Online Target Generation
Kyungjune Baek
Preprint 2026, Under Review
Lost in the Noise: A Stealthy Computation Cost Attack on 3DGS by Targeting Perceptually Vulnerable Regions
Kyungjune Baek
Preprint 2026, Under Review
Multi-dimensional Preference Alignment by Conditioning Reward Itself
Jiho Jang, Jin-Young Kim, Kyungjune Baek†, Nojun Kwak†
Preprint 2025, Under Review
Zero-to-Interaction: Generating Dynamic Videos from Synthetic State Transitions
Jiho Jang*, Jin-Young Kim*, Nojun Kwak, Kyungjune Baek†
Preprint 2025, Under Review
Measuring Bias in Text-to-Image Generative Models through Internal Representation Analysis
Hyeongmin Lee, Kyungjune Baek†
Preprint 2025, Under Review
TWLV-I: Analysis and Insights from Holistic Evaluation on Video Foundation Models
Hyungmin Lee*, Jinyoung Kim*, Kyungjune Baek*, Jihwan Kim*, and Twelve Labs Team
Technical Report 2024

Publications
(* : equal contribution. / † : corresponding author.)

Toward Stable World Models: Measuring and Addressing World Instability in Generative Environments
Soonwoo Kwon*, Jin-Young Kim*, Hyojun Go, Kyungjune Baek†
Pattern Recognition 2026 (IF=7.6)
Rethinking Direct Preference Optimization in Diffusion Models
Junyong Kang*, Seohyun Lim*, Kyungjune Baek, Hyunjung Shim†
AAAI Conference on Artificial Intelligence (AAAI) 2026 (Oral)
MomentMix Augmentation with Length-Aware DETR for Temporally Robust Moment Retrieval
Seojeong Park, Jiho Choi, Kyungjune Baek, Hyunjung Shim†
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2026
Temporal Smoothness-Aware Rate-Distortion Optimized 4D Gaussian Splatting
Hyeongmin Lee, Kyungjune Baek†
Conference on Neural Information Processing Systems (NeurIPS) 2025
Generic and Privacy-free Synthetic Data Generation for Pretraining GANs
Kyungjune Baek, Hyunjung Shim†
Conference on Neural Information Processing Systems (NeurIPS) 2022 Workshop
Learning from Better Supervision: Self-distillation for Learning with Noisy Labels
Kyungjune Baek*, Seungho Lee*, Hyunjung Shim†
International Conference on Pattern Recognition (ICPR) 2022
Logit Mixing Training for More Reliable and Accurate Prediction
Duhyeon Bang*, Kyungjune Baek*, Jiwoo Kim, Yunho Jeon, Jin-Hwa Kim, Jiwon Kim, Jongwuk Lee, Hyunjung Shim†
International Joint Conference on Artificial Intelligence (IJCAI) 2022
Commonality in Natural Images Rescues GANs: Pretraining GANs with Generic and Privacy-free Synthetic Data
Kyungjune Baek, Hyunjung Shim†
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2022
Rethinking the Truly Unsupervised Image-to-Image Translation
Kyungjune Baek, Yunjey Choi, Youngjung Uh, Jaejun Yoo, Hyunjung Shim†
IEEE/CVF International Conference on Computer Vision (ICCV) 2021
GridMix: Strong Regularization Through Local Context Mapping
Kyungjune Baek*, Duhyeon Bang*, Hyunjung Shim†
Pattern Recognition 2021 (IF=7.2)
PsyNet: Self-Supervised Approach to Object Localization Using Point Symmetric Transformation
Kyungjune Baek*, Minhyun Lee*, Hyunjung Shim†
AAAI Conference on Artificial Intelligence (AAAI) 2020
Editable Generative Adversarial Networks: Generating and Editing Faces Simultaneously
Kyungjune Baek, Duhyeon Bang, Hyunjung Shim†
Asian Conference on Computer Vision (ACCV) 2018 (Oral, Best paper Runner-up)

Projects

National Program of Excellence in Software
Sejong University
Managing student Hackerthon and Coordinated university-industry collaboration to connect software students with professionals.
Multi-domain Video Foundation Model
Twelve Labs. · Project Lead
Developing a video foundation model working on various vision and audio domains (such as RGB, depth, segmentation, and spectrogram).
Video Foundation Model and Evaluation
Twelve Labs. · Member
Developing and evaluating a video foundation model. Designing training objectives, models, and architecture. Optimizing distributed training speed.
End-to-End Document Recognition
Upstage · Member
Developing a multi-modal model to handle processes such as text and non-text detection, recognition, and named entity filtering in an end-to-end manner.
Non-text Entity Recognition in Document
Upstage · Project Lead
Developing a model to recognize non-text entities (e.g., checkbox, signature, and stamp) in a document.
Synthetic Document Generation
Upstage · Project Lead
Generating various synthetic document images for training end-to-end named entity recognition and document understanding.
AI R&D Grand Challenge for Daily Life
Yonsei University · Member
Composing a dataset of trash and training a model on the dataset for classifying the blind test set.