Our lab led by Jin-Hong Du works at the intersection of statistics, machine learning, and data-driven science. Our research combines rigorous statistical theory with modern computational methods to address problems involving causality, interpretability, distribution shifts, and complex structured data.

Our current research includes:

• Causal inference Identification, semiparametric inference, treatment-effect estimation, transportability, and causal learning under weak assumptions.

• Reliable and interpretable machine learning Feature importance, uncertainty quantification, distribution-shift robustness, and principled evaluation of machine-learning systems.

• High-dimensional and structured data Statistical methods for networks, time series, latent-factor models, functional data, and other complex dependent data.

• AI for scientific discovery Machine learning for single-cell genomics, perturbation modeling, biomedical research, and other scientific applications.

• Foundation models and intelligent systems Statistical foundations for reasoning, causal representation, out-of-distribution generalization, and verifiable evaluation of foundation models and agents.

This organization hosts:

• Research software and reproducible implementations
• Code accompanying our papers and preprints
• Simulation and benchmarking frameworks
• Tutorials, examples, and research resources
• Collaborative projects developed by lab members

Repositories will include documentation, reproducibility instructions, and licensing information whenever possible.

We aim to build methods that are:

• Statistically principled
• Computationally practical
• Transparent and reproducible
• Robust to real-world complexity
• Relevant to substantive scientific questions

We welcome collaborations across statistics, machine learning, artificial intelligence, genomics, and data-intensive scientific disciplines.

For research information, publications, and updates, visit Jin-Hong Du’s website.