Istock4

Data Science Seminar Series

The data science seminar series aims to promote research related to machine learning, computer science, statistics and their interface. We invite both internal and external speakers to present their latest cutting edge research. All staff and students are welcome to attend our seminars!

Michaelmas Term 2022

Monday 17 October 2022, 2-3pm - Caroline Uhler (MIT)

 

Caroline_Uhler2

Website 

Biography - Caroline Uhler is a Full Professor in the Department of Electrical Engineering and Computer Science and the Institute for Data, Systems, and Society at MIT. In addition, she is a core institute member at the Broad Institute, where she co-directs the Eric and Wendy Schmidt Center. She holds an MSc in mathematics, a BSc in biology, and an MEd all from the University of Zurich. She obtained her PhD in statistics from UC Berkeley in 2011 and then spent three years as an assistant professor at IST Austria before joining MIT in 2015. She is a Simons Investigator, a Sloan Research Fellow, and an elected member of the International Statistical Institute. In addition, she received an NSF Career Award, a Sofja Kovalevskaja Award from the Humboldt Foundation, and a START Award from the Austrian Science Foundation. Her research lies at the intersection of machine learning, statistics, and genomics, with a particular focus on causal inference, representation learning, and gene regulation.

Title - From Interventions to Causality using Over-Parameterized Neural Networks.

Abstract - Massive data collection holds the promise of a better understanding of complex phenomena and ultimately, of better decisions. An exciting opportunity in this regard stems from the growing availability of perturbation / intervention data (for example from drug/knockout screens in biology, advertisement, online education, etc.). In order to obtain mechanistic insights from such data, a major challenge is the development of a framework that integrates observational and interventional data and allows causal transportability, i.e., predicting the effect of unseen interventions or transporting the effect of interventions observed in one context to another. I will discuss how over-parameterized neural networks can be used for these problems. In particular, I will characterize the implicit bias of over-parameterized autoencoders and link this to causal transportability in the context of virtual drug screening.

➡️Register here

Past Seminars archive

MT 2020 -  LT 2021 (PDF)

MT 2021 - LT 2022 (PDF)