Scalable Bayesian inference for secure and reliable decision making (2025-2028)

This project aims to develop the next generation of scalable Bayesian algorithms for fitting complex statistical models with big data. Existing scalable algorithms are either approximate or restricted to simple models with a small number of parameters. This can lead to unreliable data-informed decision making. Moreover, these methods are typically reliant on processing and storing big data at a central location which presents a higher risk to the privacy of the data. This project expects to significantly expand the set of problems where scalable Bayesian learning is practical and safe. This will lead to significant benefits for practitioners looking to make reliable decisions when working with big data.