Jinhyeok Jang 

Research Interests

• Robotics, Vision-Language-Action

• AI Safety, Trustworthy AI

• Efficient Training, Acceleration

Publications

(C: Conference, J: Journal, P: Preprint, *: Equal Contribution)

2026

• [C16] Gradient-Mine Units: Scorched-Earth Strategy for Model Protection against Unauthorized Fine-Tuning
  J. Jang, J. Kim, B.O. Han
  The Fortieth Annual Conference on Neural Information Processing Systems (NeurIPS), 2026 (Accepted!)

• [C15] Not All Task Vectors Need Equal Rank: Energy-Proportional Allocation for Model Merging
  H. Cho, J. Jang (Corresponding Author)
  18th Asian Conference on Computer Vision (ACCV), 2026  (Oral)

• [C14] Trinity: Human–Agent–Environment Alignment for Embodied Task Planning  

H. Kim, H. Jeon, D. Kim, J. Jang, M. Jang

The 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP) Findings, 2026

• [C13] KARST: Knowledge-Augmented Human Action Recognition via Prompt Tuning with Structured Text for Vision-Language Models   

S. H. Nguyen, D. Wang, J. Jang, H. Seo

IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), 2026

• [C12] Learning from Oblivion: Predicting Knowledge Overflowed Weights via Retrodiction of Forgetting [paper] 

J. Jang, J. Kim, J. U. Kim

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026 (Highlight)

• [C11] IVRA: Improving Visual-Token Relations for Robot Action Policy with Training-Free Hint-Based Guidance

J. Park, K. Ranasinghe, J. Jang, C. Mata, Y. S. Jang, M. S. Ryoo

IEEE International Conference on Robotics & Automation (ICRA), 2026

2025

• [C10] PRADA: Protecting and Detecting Dataset Abuse for Open-source Medical Dataset [paper]

J. Jang*, H. Lee*, N. Navab, S. Kim

International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), 2025

• [C9] Learning to Rewind via Iterative Prediction of Past Weights for Practical Unlearning [paper]

J. Jang, J. Kim, C.H. Youn

Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2025 (Oral)

2024

• [C8] A Dual Process VLA: Efficient Robotic Manipulation Leveraging VLM  [paper]

B.O. Han, J. Kim, J. Jang (Corresponding Author)

Conference on Robot Learning Workshops (CoRL Workshop), 2024

• [C7] Rethinking Data Bias: Dataset Copyright Protection via Embedding Class-wise Hidden Bias [paper]

J. Jang, B.O. Han, J. Kim, C.H. Youn 

European Conference on Computer Vision (ECCV), 2024 

• [C6] Rethinking Peculiar Images by Diffusion Models: Revealing Local Minima’s Role [paper]

J. Jang, C.H. Youn, M. Jeon, C. Lee 

Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2024

2023

• [C5] Video-based gait analysis for assessing Alzheimer’s Disease and Dementia with Lewy Bodies [paper]  

D. Wang, C. Zouaoui, J. Jang, H. Drira, H. Seo 

International Workshop on Applications of Medical AI, 2023

• [J7] Deep emotion change detection via facial expression analysis [paper]  

B.O. Han, C.H. Yoo, H.W. Kim, J.H. Yoo, J. Jang (Corresponding Author)

Neurocomputing, 2023

• [C4] Learning to Boost Training by Periodic Nowcasting Near Future Weights [paper]  

J. Jang, W. Yun, W.H. Kim, Y. Yoon, J. Kim, J. Lee, B.O. Han 

International Conference on Machine Learning (ICML), 2023

• [J6] Virtual portraits from rotating selfies [paper]  

Y. Lee, J. Jang, S. Yang 

ETRI Journal, 2023

2022

• [C3] Real-world Validation Study of Daily Activity Detection for the Elderly [paper]  

M. Cho, J. Jang, J. Lee, M. Jang, D. Kim, J. Kim 

IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), 2022

2021

• [J5] Efficient architecture for deep neural networks with heterogeneous sensitivity [paper] 

H. Cho, J. Jang, C. Lee, S. Yang 

Neural Networks , 2021

2020

• [C2] ETRI-activity3D: A large-scale RGB-D dataset for robots to recognize daily activities of the elderly [paper]  [project]  

J. Jang, D. Kim, C. Park, M. Jang, J. Lee, J. Kim 

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2020

• [J4] Deep neural networks with a set of node-wise varying activation functions [paper] 

J. Jang, H. Cho, J. Kim, J. Lee, S. Yang 

Neural Networks , 2020

2018

• [P1] Neural Networks with Activation Networks [arXiv] 

J. Jang, J. Kim, J. Lee, S. Yang 

arXiv, 2018

• [J3] Facial attribute recognition by recurrent learning with visual fixation [paper] 

J. Jang, H. Cho, J. Kim, J. Lee, S. Yang 

IEEE Transactions on Cybernetics (TCYB), 2018

2017

• [J2] Multi-objective based spatio-temporal feature representation learning robust to expression intensity variations for facial expression recognition  [paper] 

D.H. Kim, W.J. Baddar, J. Jang, Y.M. Ro 

IEEE Transactions on Affective Computing (TAFF), 2017

• [C1] Color channel-wise recurrent learning for facial expression recognition [paper] 

J. Jang, D.H. Kim, H.I. Kim, Y.M. Ro 

IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2017 (Oral)

2016

• [J1] Modeling non-stationary asymmetric lens blur by normal sinh-arcsinh model  [paper] 

J. Jang, J.D. Yun, S. Yang 

IEEE Transactions on Image Processing (TIP), 2016

Patent

• PR20170335  평면 물체의 자세 인식 방법 및 이를 위한 장치

• PR20180201  로봇 환경에서 휴먼의 일상 행동을 인식하는 장치 및 방법

• PR20190255  로봇 환경에서 사용자에게 적응해 행동을 인식하는 장치 및 방법

• PR20200067  CCTV 환경에서 영상 해석을 통하여 사람의 이상행동을 검출하는 장치 및 방법

• PR20200260  타겟 사용자 유형에 따른 연합 학습 및 모델 성향 분석 시스템

Academic Activities

• Organization Committee: CoRL 2025

• Outstanding Reviewer: ICML 2022 

• Conference Reviewer: NeurIPS, ICML, ICLR, AAAI, CVPR, ICCV