Taekyung Kim

Research Scientist @ NAVER AI Lab

prof_me.jpg

My research focused on understanding complex dynamics and mapping them to real-world responses, ranging from linguistic responses to physical reactions.

I obtained my Ph.D. in Electrical Engineering at Korea Advanced Institute of Science and Technology (KAIST) in 2022, advised by Prof. Changick Kim. From 2016, I attended a graudate program in mathematics at KAIST for one and a half years. I obtained my B.S. in Mathematical Science at KAIST in 2016.

News

Sep. 2026 One paper is accepted at CoRL 2026.
Aug. 2026 One paper is accepted at EMNLP 2026.
Jun. 2026 One paper is accepted at ECCV 2026.
Feb. 2026 Two papers are accepted at CVPR 2026. One paper is accepted at CVPRW 2026.
Jan. 2026 Two papers are accepted at ICLR 2026. One paper is accepted at ICLRW 2026.

Selected Publications

  1. CoRL
    SPARK: Simple Post-training for Adapting pRetrained Knowledge to Robot Control
    Taekyung Kim, Jeongeun Park, Byeongho Heo, Sungjoon Choi, Sangdoo Yun, and Dongyoon Han
    Accepted to The Conference on Robot Learning & CVPRW 2026 ScaleBot, 2026
  2. EMNLP
    Beneath the Surface of Chains-of-Thought: A Mechanistic Interpretation of Reasoning Operations in LLMs
    Seogyeong Jeong, Jaehui Hwang, Dongyoon Han, Geonmo Gu, Alice Oh†, and Taekyung Kim†
    Accepted to The Conference on Empirical Methods in Natural Language Processing (EMNLP), 2026
  3. CVPR
    ORCA: Exploring Conditions for Diffusion models in Robotic Control
    Heeseong Shin, Byeongho Heo, Dongyoon Han, Seungryong Kim†, and Taekyung Kim†
    Accepted to The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026
  4. ICLR
    Map the Flow: Revealing Hidden Pathways of Information in VideoLLMs
    Minji Kim*, Taekyung Kim*, and Bohyung Han
    Accepted to The Fourteenth International Conference on Learning Representations (ICLR), 2026
  5. NeurIPS
    Token Bottleneck: One Token to Remember Dynamics
    Taekyung Kim, Dongyoon Han, Byeongho Heo, Jeongeun Park, and Sangdoo Yun
    In The Thirty-ninth Annual Conference on Neural Information Processing Systems (NeurIPS), 2025
  6. ICLR
    Morphing Tokens Draw Strong Masked Image Models
    Taekyung Kim, Byeongho Heo, and Dongyoon Han
    In The Thirteenth International Conference on Learning Representations (ICLR), 2025
  7. ECCV
    Leveraging Temporal Contextualization for Video Action Recognition
    Minji Kim, Dongyoon Han, Taekyung Kim†, and Bohyung Han†
    European Conference on Computer Vision (ECCV), 2024
  8. ECCV
    Learning with Unmasked Tokens Drives Stronger Vision Learners
    Taekyung Kim, Sanghyuk Chun, Byeongho Heo, and Dongyoon Han†
    In European Conference on Computer Vision (ECCV), 2024
  9. ICCV
    Just a Few Points are All You Need for Multi-View Stereo: A Novel Semi-Supervised Learning Method for Multi-View Stereo
    Taekyung Kim, Jaehoon Choi, Seokeon Choi, Dongki Jung, and Changick Kim
    In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), Oct 2021
  10. ECCV
    Attract, Perturb, and Explore: Learning a Feature Alignment Network for Semi-supervised Domain Adaptation
    Taekyung Kim and Changick Kim
    In European Conference on Computer Vision (ECCV), 2020
  11. CVPR
    Diversify and Match: A Domain Adaptive Representation Learning Paradigm for Object Detection
    Taekyung Kim, Minki Jeong, Seunghyeon Kim, Seokeon Choi, and Changick Kim
    In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019