|
Changhyun Choi
I am an integrated master`s & doctoral course student in the Lab for Autonomous Robotics Research(LARR) at SNU IPAI (Seoul National University Interdisciplinary Program in Artificial Intelligence), advised by Prof. H. Jin Kim.
I received my B.S. degree in Mechanical Engineering from Yonsei University(Cum Laude, ranking 7th out of 141 students).
My current research interests include aligning text-to-image generation models and GFlowNets (particularly diffusion samplers). Representative news items are highlighted below.
Colors are used to categorize different types of news: red for papers, blue for projects, green for awards, and gray for other news(e.g., experiences).
If you are interested or have any questions, please feel free to email me.
Email /
Google Scholar /
Github /
X
|
|
|
News
- 09.2025 (Since 09.2024):
- Parameter Tuning for Transferring Base Vehicle Driving Characteristics to Derivative Vehicles: We are currently conducting research with Hyundai Motor Group to develop a method that, given a vehicle with arbitrary parameters, uses MAML (a form of meta-reinforcement learning) to accurately achieve target accelerations at various speeds under specific APS conditions. This approach aims to effectively transfer the driving characteristics of a base vehicle to its derivative models.
- 03.2025~:
- Manned-Unmanned Aircraft Collaboration: We are currently conducting research with ADD (Agency for Defense Development) to achieve efficient collaboration between human pilots and unmanned aircraft using various AI methodologies.
- 03.2025 (Since 03.2024):
- AI Virtual Aircraft Air-to-Air Combat: We collaborated with KAI (Korea Aerospace Industries), employing multi-agent reinforcement learning, to increase the win rate of AI-operated virtual aircraft in air-to-air combat.
- 12.2024:
-
KSC 2024: I attended the
Korea Software Congress 2024
and presented two papers.
(Noise Generation using Noise-Selecting GFlowNet for Preference Alignment of Diffusion Models / Changhyun Choi, H. Jin Kim)
(Survey on Generative Flow Network / Changhyun Choi, H. Jin Kim)
- 11.2024 (Since 09.2023):
- Swarm Unmanned Surface Vehicle (USV) Defense Mission: We collaborated with ADD (Agency for Defense Development), using multi-agent reinforcement learning, to develop a mission where multiple friendly unmanned surface vehicles intercept and eliminate incoming enemy forces at sea before they reach their designated destination.
- 01.2024:
- Personal Project: I carried out a project that applied MCdropout to enable adaptive vehicle movement under uncertain position estimates. I used the Pure Pursuit algorithm to calculate the steering angle.
- 09.2023:
- Admission: I began my combined master’s and doctoral degree program.
- 08.2022 (Since 07.2022):
- Research Intern: I worked as a research intern at the ML Lab(POSTECH), advised by Prof. Jungseul Ok. Although the short duration limited my results, I learned how to conduct research thanks to the guidance of the professor and lab members.
- 12.2021 (Since 01.2021):
- Club: At Yonsei University, I served as the president of SBTM, an autonomous driving club. Due to COVID-19, we were unable to host many in-person events, but we organized virtual seminars and worked on various projects. Ultimately, we won the AIM (AI & IoT for Mobility) Club Contest, and a patent attorney deemed one of our ideas potentially patentable.
- 10.2019 (Since 02.2018):
- Military Service: I completed my military service at the Korea Combat Training Center (KCTC), serving as part of an opposing force to help train the Republic of Korea Army in realistic combat scenarios.
|
|