Junming Yin

I work at the intersection of machine learning, statistics, and business, focusing on the design of novel statistical models and computational algorithms for solving cutting-edge business problems. I received a Ph.D. in Computer Science and an M.A. in Statistics from UC Berkeley (advisors: Michael I. Jordan and Yun S. Song).

Research Interests:

Foundation Models, Synthetic Data Generation, Deep Generative Models
User Behavioral Dynamics Modeling, Neural Point Processes, Time Series
Nonparametric and High-Dimensional Statistics, Uncertainty Quantification, Variational Inference

Student Mentees: Jerry Luo (Ph.D. in Math; now at Google), Xi Wang (CS; now Ph.D. student at Johns Hopkins)

Editorial Board: JMLR and ACM TOPML

Email: junming.yin.work[at]gmail[dot]com

Recent Highlights:

Two demo songs written and recorded in April 2026: Everything Must End and (What) If.

When LLMs get significantly worse: A statistical approach to detect model degradations. ICLR 2026.

Mitra: Mixed synthetic priors for enhancing tabular foundation models. NeurIPS 2025.

Modeling behavioral dynamics in digital content consumption: An attention-based neural point process approach with applications in video games. Marketing Science, 44(1), 220-239, 2025.

Learning from crowdsourced multi-labeling: A variational Bayesian approach. Information Systems Research, 32(3), 752–773, 2021.

Relaxed multivariate Bernoulli distribution and its applications to deep generative models. UAI 2020.