
I’m a second-year CS Ph.D. student at Stanford, advised by Chelsea Finn and part of the IRIS lab. I am affiliated with SAIL, CRFM, and the ML Group. My research is supported by KFAS.
Previously, as alternative military service for the South Korean army, I worked as a research scientist at Kakao and AITRICS, working with Juho Lee. Before that, I completed my master’s (CS, advised by Seungjin Choi) and undergraduate (math) degrees at POSTECH.
Real-world conditions are nonstationary rather than static. My research interest is in building reliable machine learning systems that can navigate and make sound decisions in such perpetually changing environments. Here are some key questions that guide my research:
- Adaptation: How can we build systems that can quickly and robustly adapt in changing conditions?
- Understanding fine-tuning: How can we better conceptualize fine-tuning as applied in practice? What knowledge is present in foundation models, and what factors influence how much knowledge is preserved during fine-tuning?
- Underspecification: No dataset fully specifies its intended task. How can we make models recognize and represent the multitude of possible realities consistent with given data?
- De-risking errors: What strategies can we employ to handle the reality of machine learning systems generating potentially erroneous outputs?
- Usable information: How can we formalize and quantify the amount of information in a dataset that is (1) learnable by a neural network and/or is (2) pertinent to a given task?
- Better teaching: Is there a more efficient or robust way to teach machines, beyond passive observation or imitation, so that they more easily “understand” the underlying concepts?
Selected Papers
Conservative Prediction via Transductive Confidence Minimization
arXiv:2306.04974