Mincheol Kwon

Hello! I am a Master's student at Korea University, co-advised by Prof. Jinkyu Kim and Prof. Jungbeom Lee, and working in collaboration with Prof. Synho Do (Harvard Medical School) and Prof. Seunghyun Park (Soongsil University).

My research analyzes the inherent limitations of AI, aiming to make it effective and trustworthy across diverse domains.

Email / CV / Google Scholar / GitHub / LinkedIn

News

Aug. 2026	SAGE paper accepted to EMNLP 2026 Main Conference.
Apr. 2026	TruthProbe paper accepted to ICML 2026.
Mar. 2026	PinPoint paper accepted to CVPR 2026.
Jun. 2025	Joined Harvard Medical School as a Visiting Researcher.

Research

My research interests include Vision-Language Models (VLMs), World Models, and Vision-Language-Action (VLA) models.

* denotes equal contribution.
	SAGE: Semantic Attribute Graphs for Multi-Entity Visual Retrieval Yongjoo Kim, Mincheol Kwon*, Seonga Choi, Minseung Lee, Kyeong-Jin Oh, Hyunyoung Lee, Yunsu Choi, Jungbeom Lee Empirical Methods in Natural Language Processing (EMNLP Main), 2026, Budapest, Hungary. (Acceptance Rate: 15.4%)* project page / arXiv / code
	The Truth Stays in the Family: Enhancing Contextual Truthfulness via Inherited Heads in Model Lineages Miso Choi, Seonga Choi, Mincheol Kwon, Woosung Jung, Jinkyu Kim International Conference on Machine Learning (ICML), 2026 arXiv / code
	Focus, Don't Prune: Identifying Instruction-Relevant Regions for Information-Rich Image Understanding Mincheol Kwon, Minseung Lee, Seonga Choi, Miso Choi, Kyeong-Jin Oh, Hyunyoung Lee, Cheonyoung Park, Yongho Song, Seunghyun Park, Jinkyu Kim The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026, Denver CO, USA. (Acceptance Rate: 25.4%) project page / arXiv / code
	Watermarking for Factuality: Guiding Vision-Language Models Toward Truth via Tri-layer Contrastive Decoding Kyungryul Back, Seongbeom Park, Milim Kim, Mincheol Kwon, SangHyeok Lee, Hyunyoung Lee, Junhee Cho, Seunghyun Park, Jinkyu Kim Empirical Methods in Natural Language Processing (EMNLP Findings), 2025 arXiv / code
	A Dynamic-Selection-Based, Retrieval-Augmented Generation Framework: Enhancing Multi-Document Question-Answering for Commercial Applications Mincheol Kwon, Jimin Bang, Seyoung Hwang, Junghoon Jang, Woosin Lee Electronics*, vol.14, MDPI, 2025 paper / code

Design based on Jon Barron's website.