Yo Joong “YJ” Choe

I am an Assistant Professor of Decision Sciences at INSEAD (resident faculty in Singapore). I am an interdisciplinary researcher with interests in statistics, machine learning, and large language models. My recent research spans two areas:

Statistical methods for sequential anytime-valid inference (SAVI), in particular e-values:
testing for upsides by “betting on bets”; combining evidence under information asymmetry; comparing sequential forecasters; feature-aware forecast audits
Representation geometry in large language models (LLMs): linear representations; causal notions of interpretability; model steering; softmax information geometry (#1, #2, #3)

I hold a joint Ph.D. in Statistics and Machine Learning from Carnegie Mellon University, where I was advised by Aaditya Ramdas. From 2023 to 2025, I was a DSI Faraco Family Postdoctoral Fellow at the University of Chicago, where I worked with Victor Veitch.

Recent news and upcoming events (details):
Revised working paper titled “betting on bets” is now on arXiv. (paper page)
New working paper on feature-aware forecast audits is now on arXiv. (link)
New paper on softmax geometry in LLMs is accepted to ICML. (paper page)
The “combining evidence” paper is now published in JRSS-B. (paper page)
Summer: presented at IMS NRC Asia, INFORMS ADA, SAVI, and ICML. Links to
slides (betting on bets), poster (betting on bets), and slides (comparing forecasters by betting).
Fall: co-organizing a NeurIPS workshop on e-values (Paris); organizing a session on e-values at INFORMS (SF); teaching Data Analysis on our Europe campus (Fontainebleau).

CV Papers Teaching Software News Blog Academic Bio