Research Interests
I develop Bayesian methods that are computationally efficient and supported by theoretical guarantees. My recent work focuses on flexible models and scalable algorithms for large, complex data sets in genomics, geostatistics, and biomedicine. Because current approaches often struggle with both size and complexity, I design tractable methods that draw on machine learning and asymptotic Bayesian statistics.
My current research has three themes:
- Scalable sampling algorithms using divide-and-conquer and asynchronous computation for Bayesian inference
- Parametric and nonparametric models for high-dimensional data
- Array-variate models for multimodal neuroscience data, including local field potentials, imaging, and gene expression
I am also exploring causal inference, deep learning, and large language models.
Code for many manuscripts is available on GitHub (@blayes).
Publications
- By citations: Google Scholar
- By year: Google Scholar
Funding
My research has been supported by the Office of Naval Research, the National Science Foundation Division of Mathematical Sciences, and the National Institute of Mental Health. Current support includes NSF DMS-2506058 (Co-PI) and NIH R01 OD039332 (Co-Investigator with Rainbo Hultman and Hanna Stevens as MPIs).