Variable Selection and Prioritization in Bayesian Machine Learning Methods

A consistent theme of the work done in my lab group is to take modern computational approaches and develop theory that enable their interpretations to be related back to classical genomic principles. The central aim of this talk is to address variable selection questions in nonlinear and nonparametric regression. Motivated by statistical genetics, where nonlinear interactions and non-additive variation are of particular interest, we introduce a novel, interpretable, and computationally efficient way to summarize the relative importance of predictor variables. Methodologically, we present flexible and scalable classes of Bayesian feedforward models which provide interpretable probabilistic summaries such as posterior inclusion probabilities and credible sets for association mapping tasks in high-dimensional studies. We illustrate the benefits of our methods over state-of-the-art linear approaches using extensive simulations. We also demonstrate the ability of these methods to recover both novel and previously discovered genomic associations using real human complex traits from the Wellcome Trust Case Control Consortium (WTCCC), the Framingham Heart Study, and the UK Biobank.

This will be a virtual event. It will be held in 116 Old Chemistry and it will be on zoom.
Join Zoom Meeting
https://duke.zoom.us/j/92397382385?pwd=NW5aTmZSWUpzOCtneTJDaFZFV0l4dz09
Meeting ID: 923 9738 2385
Passcode: 425966