Machine Learning & Causality: Building Efficient, Reliable Models for Decision-Making - Maggie Makar

Computer Science Seminar Series
February 16, 2021
“Machine Learning & Causality: Building Efficient, Reliable Models for Decision-Making”
Maggie Makar, Massachusetts Institute of Technology
Increasingly, practitioners are turning to machine learning to build causal models and predictive models that perform well under distribution shifts. However, current techniques for causal inference typically rely on having access to large amounts of data, limiting their applicability to data-constrained settings. In addition, empirical evidence has shown that most predictive models are insufficiently robust with respect to shifts at test time. In this talk, Maggie Makar will present her work on building novel techniques to address both of these problems. Much of the causal literature focuses on learning accurate individual treatment effects, which can be complex and hard to estimate from small samples. However, it is often sufficient for the decision-maker to have estimates of upper and lower bounds on the potential outcomes of decision alternatives to assess risks and benefits. Makar will show that, in such cases, we can improve sample efficiency by estimating simple functions that bound these outcomes instead of estimating their conditional expectations. She will present a novel algorithm that leverages these theoretical insights. Makar will also talk about approaches to deal with distribution shifts using causal knowledge and auxiliary data. She will discuss how distribution shifts arise when training models to predict contagious infections in the presence of asymptomatic carriers. She will present a causally-motivated regularization scheme that enables prediction of the true infection state with high accuracy even if the training data is collected under biased test administration.
Maggie Makar is a PhD student at the Massachusetts Institute of Technology (MIT) Computer Science and Artificial Intelligence Laboratory. While at MIT, Makar interned at Microsoft Research and Google Brain. Prior to MIT, she worked at Brigham and Women’s Hospital studying end-of-life care. Makar's work has appeared at the International Conference on Machine Learning, the Association for the Advancement of Artificial Intelligence Conference on Artificial Intelligence, and the Joint Statistical Meetings and in the Journal of the American Medical Association, Health Affairs, and Epidemiology, among other venues and publications. She received a BSc in mathematics and economics from the University of Massachusetts, Amherst.