Wooseong Jeong

Ph.D. Candidate at KAIST, advised by Prof. Kuk-Jin Yoon at the Visual Intelligence Lab (VILAB).

prof_pic.jpg
stk14570@kaist.ac.kr

I am a Ph.D. candidate at KAIST working on foundation model optimization and adaptation, with applications to embodied intelligence. My research focuses on multi-task optimization, model merging, and test-time adaptation, and I am currently extending this work to vision-language-action models.

Feel free to explore my publications, and reach out if you’d like to collaborate or have any questions.

Research Interests

  • Foundation Model Optimization and Adaptation
    • Multi-task, multi-objective, and non-convex optimization
    • Parameter-efficient adaptation, model merging, and knowledge distillation (including on-policy distillation)
    • Test-time adaptation and in-context learning
  • Embodied Intelligence
    • Vision-language-action models and uncertainty-guided reasoning
    • Robust perception and motion planning for autonomous systems

Honors and Awards

Recent Publications

2026

  1. yang2026triggering.png
    Triggering Generalist Reasoning via Predictive Uncertainty for Dual-System VLA
    Hyemin Yang*, Wooseong Jeong*, Giwon Lee*, and Kuk-Jin Yoon
    In Conference on Neural Information Processing Systems (NeurIPS), 2026
    arXiv Video Website
  2. jeong2026preference.png
    Preference-Aligned LoRA Merging: Preserving Subspace Coverage and Addressing Directional Anisotropy
    Wooseong Jeong*, Wonyoung Lee*, and Kuk-Jin Yoon
    In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026
    arXiv Video Code Poster
  3. lee2026labelfree.png
    Label-Free Cross-Task LoRA Merging with Null-Space Compression
    Wonyoung Lee*, Wooseong Jeong*, and Kuk-Jin Yoon
    In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026
    arXiv Video Code Poster
  4. jeong2026stabilizing.png
    Stabilizing Multi-Task Latent Spaces: Recursive Refinement with Coordinators in Partially Labeled Learning
    Wooseong Jeong*, Jegyeong Cho*, Youngho Yoon*, Jaeyoung Lee, and Kuk-Jin Yoon
    IEEE Access, 2026
    PDF

2025

  1. lee2025interaction.png
    Interaction-Merged Motion Planning: Effectively Leveraging Diverse Motion Datasets for Robust Planning
    Giwon Lee*, Wooseong Jeong*, Daehee Park, Jaewoo Jeong, and Kuk-Jin Yoon
    In The International Conference on Computer Vision (ICCV), 2025
    Highlight
    Qualcomm Fellowship Finalist arXiv Code Poster
  2. jeong2025resolving.png
    Resolving Token-Space Gradient Conflicts: Token Space Manipulation for Transformer-Based Multi-Task Learning
    Wooseong Jeong and Kuk-Jin Yoon
    In The International Conference on Computer Vision (ICCV), 2025
    arXiv Code Poster
  3. jeong2025synchronizing.png
    Synchronizing Task Behavior: Aligning Multiple Tasks during Test-Time Training
    Wooseong Jeong*, Jegyeong Cho*, Youngho Yoon*, and Kuk-Jin Yoon
    In The International Conference on Computer Vision (ICCV), 2025
    arXiv Code Poster
  4. kim2025dc.png
    DC-TTA: Divide-and-Conquer Framework for Test-Time Adaptation of Interactive Segmentation
    Jihun Kim*, Hoyong Kwon*, Hyeokjun Kweon*, Wooseong Jeong, and Kuk-Jin Yoon
    In The International Conference on Computer Vision (ICCV), 2025
    arXiv Video Code
  5. jeong2025multi.png
    Multi-View 3D Scene Abstraction From Drone-Captured RGB Images
    Wooseong Jeong*, Jihun Kim*, Hyeokjun Kweon*, and Kuk-Jin Yoon
    IEEE Access, 2025
    PDF
  6. jeong2025selective.png
    Selective Task Group Updates for Multi-Task Optimization
    Wooseong Jeong and Kuk-Jin Yoon
    In The International Conference on Learning Representations (ICLR), 2025
    arXiv Code Poster

2024

  1. jeong2024quantifying.png
    Quantifying Task Priority for Multi-Task Optimization
    Wooseong Jeong and Kuk-Jin Yoon
    In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024
    arXiv Code Poster