Yuxin Chen
Assistant Professor of Computer Science
University of Chicago
My research studies how learning systems should acquire information and make decisions when data, experiments, or interactions are costly. I develop methods in active learning, Bayesian optimization, and reinforcement learning, with a particular interest in jointly learning representations and acquisition strategies. This work spans methodological foundations and scientific applications, including autonomous experimentation in physics and materials science, as well as emerging questions in data quality, scaling laws, and optimization for foundation models.
At UChicago, I lead the Interactive Learning Systems group.
- Faculty Co-Lead, Science Labs that Only AI Can Build
UChicago AI Pillar - AI/Workflow Lead, PoLARIS
NSF Programmable Cloud Laboratory
Research areas
Active Learning & Representation Learning
Active data acquisition, machine teaching, learned acquisition functions, active representation learning.
Bayesian Optimization & Adaptive Experimental Design
Bayesian optimization, multi-fidelity optimization, constrained and multi-objective design, adaptive experimentation.
Reinforcement Learning & Sequential Decision-Making
Policy improvement, curriculum learning, model-based RL, and continual learning.
Machine Learning for Scientific Discovery
Scientific experimentation, materials, physics, PDEs, autonomous experimental systems.
Selected work
-
Direct Regret Optimization in Bayesian Optimization
ICML ExAI 2025
-
Beyond Reverse KL: Generalizing Direct Preference Optimization with Diverse Divergence Constraints
ICLR 2024 · Spotlight presentation
The group
The Interactive Learning Systems (ILS) group is a machine learning research group at the University of Chicago. Our students and postdocs work on active learning, Bayesian optimization, reinforcement learning, and machine learning for scientific discovery.
Recent news