My research focuses on developing reliable and efficient AI solutions for scientific and healthcare applications. I am particularly interested in identifying the statistical structure of data and using it to guide the development of efficient, reliable learning systems. My work has produced AI solutions across space science, biomedical applications, and experimental design.
Research interests: Reliable and Efficient AI, Experimental Design, Hypothesis Testing.
I am seeking opportunities as an Assistant Professor, Research Scientist, or Machine Learning Engineer. Please do not hesitate to contact me if you think I am a good fit!
Research Directions
Reliable and Efficient AI for Scientific and Healthcare Applications
Many scientific and healthcare applications involve limited, heterogeneous, noisy, or costly-to-annotate data. I develop reliable and data-efficient AI methods for these settings, including approaches for robust learning, efficient use of annotations, and statistically valid monitoring after deployment. My work aims to ensure that AI systems capture meaningful scientific and clinical signals rather than nuisance variation.
Selected publications
- Weizhi Li, Natalie Klein, Brendan Gifford, Elizabeth Sklute, Carey Legett, and Samuel Clegg. "Regularization via f-Divergence: An Application to Multi-Oxide Spectroscopic Analysis." JGR: Machine Learning and Computation, 2026.
- Pavel Dolin, Weizhi Li, Gautam Dasarathy, and Visar Berisha. "Statistically Valid Post-Deployment Monitoring Should Be Standard for AI-Based Digital Health." NeurIPS, 2025.
- Weizhi Li, Gautam Dasarathy, and Visar Berisha. "Regularization via Structural Label Smoothing." AISTATS, 2020.
- Weizhi Li, Gautam Dasarathy, Karthikeyan Natesan Ramamurthy, and Visar Berisha. "Finding the Homology of Decision Boundaries with Active Learning." NeurIPS, 2020.
- Weizhi Li, Xiaoning Qian, and Jim Ji. "Noise-Tolerant Deep Learning for Histopathological Image Segmentation." ICIP, 2017.
Accelerating Experimental Pipelines Using AI
Experiments support critical decisions across many domains: product teams use A/B tests to assess new features, while clinical researchers conduct trials to evaluate treatments. I develop AI-enabled experimental pipelines that reduce the time and data required to reach reliable conclusions while maintaining rigorous control of false discoveries.
Selected publications
- Weizhi Li, Gautam Dasarathy, and Visar Berisha. "Matched-Pair Experimental Design with Active Learning." arXiv preprint, 2025.
- Weizhi Li, Visar Berisha, and Gautam Dasarathy. "Advanced Tutorial: Label-Efficient Two-Sample Tests." WSC, 2024.
- Weizhi Li, Prad Kadambi, Pouria Saidi, Karthikeyan Natesan Ramamurthy, Gautam Dasarathy, and Visar Berisha. "Active Sequential Two-Sample Testing." TMLR, 2023.
- Weizhi Li, Gautam Dasarathy, Karthikeyan Natesan Ramamurthy, and Visar Berisha. "A Label-Efficient Two-Sample Test." UAI, 2022.
Affiliations
Texas A&M University
Selected Talks
- "Label-Efficient Two-Sample Tests", Statistics Seminar, University of New Mexico, invited talk, 2025.
- "Label-Efficient Two-Sample Tests", Winter Simulation Conference, tutorial talk, 2024.
- "AI-Enhanced Exploration for Planetary Science", Applied Machine Learning Summer School, Los Alamos National Laboratory, invited talk, 2024.
- "Designing Two-Sample Tests with Local Information", LIONS Seminar Series, Arizona State University, invited talk, 2023.
