Causal Data Science: A general framework for data fusion and causal inference

Monday, April 1, 2019
11:40 AM - 12:40 PM
Department of Computer Science, 500 W. 120th St., New York, New York 10027
Room/Area: 451
Add to Calendar

Link added to clipboard:

https://events.columbia.edu/cal/event/eventView.do?b=de&calPath=%2Fpublic%2Fcals%2FMainCal&guid=CAL-00bb9e25-69b90bcd-0169-bc8dde85-00002c34events@columbia.edu&recurrenceId=
Computer Science Distinguished Lecture Series
Elias Bareinboim

Abstract:
Causal inference is usually dichotomized into two categories, experimental (Fisher, Cox, Cochran) and observational (Neyman, Rubin, Robins, Dawid, Pearl) which, by and large, have evolved and been studied independently. A wide range of problems faced by the current generation of empirical scientists is more demanding. Experimental and observational studies are but two extremes of a rich spectrum of research designs that generate the bulk of the data available in practical, large-scale situations. In typical medical explorations, for example, data from multiple observational and experimental studies are collected from distinct locations, different sampling conditions, and heterogeneous populations. Piecing together these data sources presents a tremendous opportunity to data scientists since the knowledge conveyed by the combined data would not be attainable from any individual source alone.

However, the biases that emerge in heterogeneous environments require a new set of principles and analytical tools. Some of these biases, including confounding, sampling selection, and cross-population (i.e., external validity) biases, have been addressed in isolation, largely in restricted parametric models. In this talk, I will present a general, non-parametric framework for handling these biases and, ultimately, a theoretical solution to the data fusion problem in causal inference tasks. I will further outline the connections of this theory to current challenges in AI and Machine Learning, including fairness analysis, explainability, and reinforcement learning. I’ll end the talk with some reflections on where we are now in the grand scheme of automating the empirical sciences, a project that I call "Causal Data Science.”

Bio:
Elias Bareinboim is an assistant professor in the Departments of Computer Science and Statistics at Purdue University. His research focuses on causal and counterfactual inference and their applications in data-driven fields. His work was the first to propose a general solution to the problem of “data-fusion” and provides practical methods for combining datasets generated under different experimental conditions. More recently, Bareinboim has been interested in the intersection of causal inference with reinforcement learning and fairness analysis. He received a Ph.D. in Computer Science from UCLA advised by Judea Pearl. Bareinboim’s recognitions include NSF CAREER Award, IEEE AI’s 10 to Watch, the Dan David Prize Scholarship, the 2014 AAAI Outstanding Paper Award, and the 2018 UAI Best Student Paper Award.
Event Contact Information:
Daniel Hsu
[email protected]
LOCATION:
  • Morningside
TYPE:
  • Lecture
CATEGORY:
  • Computer Science
EVENTS OPEN TO:
  • Alumni
  • Faculty
  • Family-friendly
  • Graduate Students
  • Postdocs
  • Prospective Students
  • Public
  • Staff
  • Students
  • Trainees
BACK TO EVENTS

Date Navigation Widget

Filter By

Subscribe Export Options

Getting to Columbia

Other Calendars

Guests With Disabilities