Beyond Tech: Machine learning in science & policy

Seminar with David Dunson, Arts and Sciences Distinguished Professor, Departments of Statistical Science, Mathematics, and Electrical & Computer Engineering, Duke University

Beyond Tech: Machine learning in science & policy


Time:14:00 – 15:00

Date: 22 May

You must register to attend.

Applications in the tech industry have driven much of the progress in machine learning in recent years.  Most tech applications are fundamentally different in structure than applications in other fields, such as science and policy making.  Hence, algorithms that can be highly successful in tech can be dramatically unsuccessful in other domains.  In this talk, I will provide a brief review of the types of application areas in which widely popular machine learning algorithms (e.g., deep learning) perform well and will highlight key differences between these settings & other areas.  I will argue that there can be disastrous results in naively applying off-the-shelf ML algorithms in areas including criminal justice (e.g, automating sentencing and bail), science (e.g, neuroscience), policy (e.g., regulating chemical exposures), and health decision making.  Instead, one needs to carefully develop targeted methods that deal with crucial issues of selection bias, uncertainty quantification, limited training data, and complex/high-dimensional observations.  As illustration, I focus in more detail on two problems: (1) removing the influence of a sensitive variable (e.g, race/ethnicity) to obtain a fair predictive algorithm (also related to causal inference and privacy); and (2) obtaining interpretable predictive models of human traits based on an individual’s brain connection structure.


David Dunson is Arts and Sciences Distinguished Professor of Statistical Science, Mathematics, and Electrical & Computer Engineering at Duke University.

His research focuses on Bayesian statistical theory and methods motivated by high-dimensional and complex applications.  A particular emphasis is on dimensionality reduction, scalable inference algorithms, latent factor models, and nonparametric approaches, particularly for high-dimensional, dynamic and multimodal data, including images, functions, shapes and other complex objects.

His work involves inter-disciplinary thinking at the intersection of statistics, mathematics and computer science. Motivation comes from applications in epidemiology, environmental health, neurosciences, genetics, fertility and other settings (music, fine arts, humanities).

Dr. Dunson is a fellow of the American Statistical Association and of the Institute of Mathematical Statistics.  He is winner of the 2007 Mortimer Spiegelman Award for the top public health statistician under 41, the 2010 Myrto Lefkopoulou Distinguished Lectureship at Harvard University, the 2010 COPSS Presidents’ Award for the top statistician under 41, and the 2012 Youden Award for interlaboratory testing methods.

Sponsored by the Lloyd’s Register Foundation Programme on Data-Centric Engineering.