AI-accelerated Bayesian methods
We develop fast and scalable Bayesian methods for parameter estimation, model comparison, and data acquisition.
Everything we do is Bayesian. We teach neural networks how to do statistics. This lets us build better models of our world.
We develop fast and scalable Bayesian methods for parameter estimation, model comparison, and data acquisition.
We craft models that turn (big) data into mechanistic insight across cognition, neuroscience, biology, and engineering.
We gratefully acknowledge the generous support of the NSF and NIH in funding our research.
Amortizing Cognitive Modeling
We are developing a framework that will allow researchers to fit millions of cognitive models. The project's focus on robustness and benchmarking aims to strengthen the emerging infrastructure for model-based analysis in the behavioral sciences.
Model-based fNIRS analysis with AI
BayesNIRS advances model-based fNIRS through simulation, generative AI, and Bayesian inference. The program brings together 3D in silico neuroimaging, real-time model calibration, and robust data interpretation for demanding applications such as surgical training.
A representative sample of our recent research outcomes. Visit Google Scholar for all of our preprints and publications.