About Me
I am a Ph.D. candidate at KAIST NMSL, under the guidance of Prof. Sung-Ju Lee, with close collaboration with Prof. Jinwoo Shin. I earned my M.S. in Electrical Engineering from KAIST, also under the supervision of Prof. Sung-Ju Lee. I completed my B.S. in Computing at KAIST, graduating Magna Cum Laude.
My research interests include end-user machine learning, including mobile AI, adaptation, and personalization. I am currently focusing on overcoming failures in test-time adaptation. My recent work includes “BiTTA” (ICML 2025), “AETTA” (CVPR 2024), and “SoTTA” (NeurIPS 2023), which focus on (1) adaptation with user guidance, (2) estimating adaptation accuracy without labeled data, and (3) adaptation with challenging noisy data streams, emphasizing the robustness and reliability of test-time adaptation.
Latest News
Jul 2025 (expected) Research intern at Microsoft Research Asia (Advisor: Lili Qiu).
May 2025 First-author paper (BiTTA) accepted at ICML ‘25 (Acceptance rate: 26.9%).
May 2025 First-author paper (CrashSniffer) accepted at MobiSys Workshop (EnvSys) ‘25.
May 2025 Invited talk at UNIST UAI lab.
Feb 2024 First-author paper (AETTA) accepted at CVPR ‘24 (Acceptance rate: 23.6%).
Sep 2023 Co-first-author paper (SoTTA) accepted at NeurIPS ‘23 (Acceptance rate: 26.1%).
Jun 2022 First-author paper accepted at CVPR Workshop (CVPM) ‘22.
Jun 2022 Demo at MobiSys ‘22.
Oct 2020 Selected as a scholarship recipient of the Kwon Yong-se Scholarship (KAIST EE).
Education
Ph.D. Candidate, KAIST
Ph.D. Candidate in School of Electrical Engineering (Advisor: Prof. Sung-Ju Lee)
2022.08 - Present
2020.08 - 2022.08
B.S., KAIST
B.S. in School of Computing
- GPA 3.96/4.30 (Magna Cum Laude), Advanced Major, Certificate in AI
2016.02 - 2020.08
Conference Publications
Taeckyung Lee, Sorn Chottananurak, Junsu Kim, Jinwoo Shin, Taesik Gong^, Sung-Ju Lee^
ICML 2025 (^: Equal advising. Acceptance rate: 26.9%)
"AETTA: Label-Free Accuracy Estimation for Test-Time Adaptation"
[ Project page ] [ Paper ] [ Code ] [ Video ]
Taeckyung Lee, Sorn Chottananurak, Taesik Gong, Sung-Ju Lee
CVPR 2024 (Acceptance rate: 23.6%)
"SoTTA: Robust Test-Time Adaptation on Noisy Data Streams"
[ Project page ] [ Paper ] [ Code ] [ Video ]
Taesik Gong*, Yewon Kim*, Taeckyung Lee*, Sorn Chottananurak, Sung-Ju Lee
NeurIPS 2023 (* Co-first author. Acceptance rate: 26.1%)
Workshop Publications
"CrashSniffer: UWB-Based Anchor-Free Pedestrian Collision Prediction for Personal Mobility Vehicles"
Taeckyung Lee, Juseung Lee, Ryuhaerang Choi, Seungjoo Lee, Hyeongheon Cha, Hyungjun Yoon, Song Min Kim, Sangwook Bak, Sung-Ju Lee.
MobiSys Workshops (EnvSys) 2025 (Oral presentation, peer-reviewed, archived)
Taeckyung Lee, Dain Kim, Sooyoung Park, Dongwhi Kim, Sung-Ju Lee
CVPR Workshops (CVPM) 2022 (Oral presentation, peer-reviewed, archived)
- Also appeared at MobiSys 2022 Demo
Work Experiences
2025.07 - 2025.12 (expected)
Networking and Mobile Systems Lab, KAIST
Undergraduate Researcher
2020.03 - 2020.08
Deepscent
Software Engineer Intern
2019.12 - 2020.02
Advanced Networking Lab, KAIST
Undergraduate Researcher
2019.03 - 2019.12
Awards
Kwon Young-se Scholarship
School of EE, KAIST (Top-2 among EE grad school applicants in 2020 Fall)
2020.10.12
The National Scholarship for Science and Engineering
Ministry of Science and ICT, Korean Government (In acknowledgment of a high GPA)
2018.08.10
Professional Services
Conference Reviewer
- NeurIPS ‘24, ‘25
- ICCV ‘25
- ICML ‘25
- CVPR ‘24, ‘25
- ICLR ‘25
Workshop Reviewer
- 2nd Workshop on Test-Time Adaptation (PUT) @ ICML ‘25
Teaching Experiences
Teaching Assistant, KAIST
- EE209 Programming Structures for EE (Spring 2025 (Head TA), Fall 2024, Spring 2023 (Head TA), Fall 2021 (Head TA))
- EE323 Computer Networks (Spring 2024, Spring 2022, Spring 2021)
- EE415 Operating Systems and System Programming for EE (Fall 2023)
- EE595 Mobile Computing, Sensing, Learning, and Interactions (Fall 2022)
- CS496 Immersion Camp (Summer 2019)
Patents
UWB-based Pedestrian Collision Avoidance System for Personal Mobility Devices
Sung-Ju Lee, Taeckyung Lee, Hyungjun Yoon, Seungjoo Lee, Hyeongheon Cha
Korea Patent, Filing date: 2025.02.12, No.10-2025-0018095
Time-Aware Reinforcement-Learning Method for Geographic Ad Hoc Routing with Pre-defined Mobility
Jaeha Ahn, Sung-Ju Lee, Taesik Gong, Hyungjun Yoon, Taeckyung Lee
Korea Patent; Date of patent: 2024.12.02, No.10-2738891, Filing date: 2022.11.11, No.10-2022-0150662
Reinforcement Learning for Ad-hoc Routing Using Pre-defined Mobility
Sung-Ju Lee, Taesik Gong, Hyungjun Yoon, Taeckyung Lee, Jaeha Ahn
Korea Patent; Date of patent: 2024.04.22, No.10-2660836, Filing date: 2022.03.21, No.10-2022-0034824
Miscellaneous
E-Sports
As a previous enthusiastic e-sports player, I am also interested in human factors in e-sports.
Analysis on Personality and In-Game Factors in League of Legends (Link)
- I conducted an in-depth analysis of the personalities and in-game factors of League of Legends players through a collaborative survey with the Professional Spectator YouTube channel.
Few records as a player:
- Top 0.1% Player (Master tier) in KR Server, League of Legends (2020)
- Champion in Daejeon Metropolitan City, International e-Sports Festival (IEF), League of Legends (2019)
- Top-500 Player in KR Server, Overwatch (2017)