Youngkyu Hong

profile photo

I am a senior undergraduate student at KAIST School of Computing and student researcher at IIDS.

Previously, I worked at Hyperconnect as a Machine Learning Engineer and Backend Engineer. At Hyperconnect, I developed ML models that enhance Azar and Hakuna's user experience, million download apps. There, I experienced existing limitations of the current state of machine learning algorithms, therefore, I decided to contribute to breaking those walls.

I believe that artificial intelligence will enable more people to take advantage of the benefits that are now only available to a few, just as the World Wide Web has changed the world. For this, I'm focusing on the challenges we have in using AI more responsibly and safely.

Email  /  CV  /  Google Scholar  /  Github


I'm interested in the theory-inspired algorithms for handling long-tail, unseen (adversarial or out-of-distribution) data possibly under resource constraints. I'm also working on causal analysis of machine learning algorithms.

(*: equal contribution)

Self-Diagnosing GAN: Diagnosing Underrepresented Samples in Generative Adversarial Networks
Jinhee Lee*, Haeri Kim*, Youngkyu Hong*, Hye Won Chung
In submission

Detect underrepresented samples in GAN training via learning dynamics analysis and emphasize them.

Disentangling Label Distribution for Long-tailed Visual Recognition
Youngkyu Hong*, Seungju Han*, Kwanghee Choi*, Seokjun Seo, Beomsu Kim, Buru Chang
CVPR, 2021

Disentangling label distribution during the training phase helps inference on the label distribution shifted dataset.

Honors and Awards
  • KFAS Undergraduate Scholarship (2018-2021)
  • Dean’s List (Spring 2016, Spring 2017, Fall 2017, and Spring 2018)
  • KAIST Presidential Fellowship (2016-2021)

Template modified from this.