Researcher on Trustworthy/Responsible AI.
Love to learn new things.
Love to play the violin.
Biography
I am a Ph.D. student at the Data Intelligence Lab of KAIST EE and advised by Professor Steven Euijong Whang. I received B.S. in electrical engineering from KAIST in 2018. I am a recipient of the Microsoft Research PhD Fellowship 2022.
My main research area is Trustworthy/Responsible AI, which includes fairness, robustness, interpretability, and transparency. I believe that these aspects will be indispensable in future machine learning systems. Among the trustworthiness aspects, I am especially interested in building fair AI systems that do not have demographic or individual disparities. To this end, I currently focus on lowering the barrier of fair AI development, achieving multiple aspects of trustworthiness, and building human-centered fair AI systems.
I was a research intern at NVIDIA Research in 2022, working with great mentors Weili Nie, De-An Huang, Arash Vahdat, and Anima Anandkumar.
Publications
Conference Papers
Improving Fair Training under Correlation Shifts [Paper]
Y. Roh, K. Lee, S. E. Whang, and C. Suh
ArXiv 2023 (Preprint)Sample Selection for Fair and Robust Training [Paper / Talk / Slides / Code]
Y. Roh, K. Lee, S. E. Whang, and C. Suh
NeurIPS 2021Machine Learning Robustness, Fairness, and their Convergence (Tutorial) [Paper / Talk / Slides]
J. Lee, Y. Roh, H. Song, and S. E. Whang
ACM SIGKDD 2021FairBatch: Batch Selection for Model Fairness [Paper / Talk / Slides / Code]
Y. Roh, K. Lee, S. E. Whang, and C. Suh
ICLR 2021Inspector Gadget: A Data Programming-Based Labeling System for Industrial Images [Paper]
G. Heo, Y. Roh, S. Hwang, D. Lee, and S. E. Whang
VLDB 2021FR-Train: A Mutual Information-Based Approach to Fair and Robust Training [Paper / Talk / Slides / Code / KAIST Breakthroughs]
Y. Roh, K. Lee, S. E. Whang, and C. Suh
ICML 2020
Journal Papers
Data Collection and Quality Challenges in Deep Learning: A Data-Centric AI Perspective [Paper]
S. E. Whang, Y. Roh, H. Song, and J. Lee
VLDB Journal 2022Responsible AI Challenges in End-to-end Machine Learning [Paper]
S. E. Whang, K. Tae, Y. Roh, and G. Heo
IEEE Data Engineering Bulletin 2021A Survey on Data Collection for Machine Learning: A Big Data - AI Integration Perspective [Paper]
Y. Roh, G. Heo, and S. E. Whang
IEEE TKDE 2021 (Early Access from 2019)
Workshop Papers
Data Cleaning for Accurate, Fair, and Robust Models: A Big Data - AI Integration Approach [Paper]
K. Tae, Y. Roh, Y. Oh, H. Kim, S. E. Whang
DEEM @ ACM SIGMOD 2019
Experience
NVIDIA, Santa Clara, CA (Virtual) | Jun. 2022 - Dec. 2022
Research Intern
Mentors: Weili Nie, De-An Huang, Arash Vahdat, and Anima Anandkumar
Education
Ph.D. (Integrated with Master) in Electrical Engineering, KAIST | Sept. 2019 - Present
Master in Electrical Engineering, KAIST | Sept. 2018 - Aug. 2019
B.S. in Electrical Engineering & Minor in Intellectual Property, KAIST | Feb. 2014 - Aug. 2018 (Summa Cum Laude )
Korea Science Academy | Feb. 2011 - Feb. 2014
Honors
Microsoft Research PhD Fellowship | 2022
Best Research Achievement Award, KAIST EE | Spring 2022
Research Breakthroughs, KAIST | Spring 2021
The Qualcomm Innovation Fellowship | Dec. 2020
Best TA Award, KAIST | Jun. 2020
Department Honors (Top-3) Scholarship, KAIST | Spring 2017
Dean's List, KAIST | Spring 2016
National Science & Engineering Scholarship | Spring 2016 - Spring 2018
Talks
Machine Learning Fairness and its Convergence with Robustness
Tutorial @ IEEE BigComp Conference | Feb. 2023
Machine Learning Robustness, Fairness, and their Convergence
Tutorial @ ACM SIGKDD Conference | Aug. 2021
Responsible AI Techniques for Model Training
TechTalk to the TensorFlow team @ Google Korea | Feb. 2020
CONtact
yuji(dot)roh(at)kaist(dot)ac(dot)kr