Causal Discovery: Challenges and Opportunities
- Friday 17 February 2023
Niels Bohrweg 1
2333 CA Leiden
A primary goal in empirical science is to discover causal relationships among a set of variables in various natural or social phenomena. In machine learning, manifestation of causality in agents equipped with a causal model of the environment could be a leap forward towards truly intelligent decision-making.
In general, using observational data, only a partial recovery of causal relationships among the variables is feasible. In contrast, if we were able to perform adequate interventions in the system, then the causal structure could be fully recovered. However, in many applications, performing interventions might be too costly or time-consuming. In this talk, I will first present the problem of experiment design to learn as many causal relationships as possible with a limited budget of interventions. Then, I will discuss an approximation algorithm for the problem of budgeted experiment design, achieving remarkable performance in real-world applications such as recovering causal structures in gene-regulatory networks. Next, I will discuss some further improvements to reduce the computational complexity of experiment design algorithms. Finally, I will conclude with my research vision towards devising causal AI systems, addressing both algorithmic and statistical challenges.
About Saber Salehkaleybar
Saber Salehkaleybar is a Scientific Collaborator in the School of Computer and Communication Sciences (IC) and College of Management Technology (CDM) at Ecole Polytechnique Federale de Lausanne (EPFL). He received B.Sc., M.Sc. and Ph.D. degrees in Electrical Engineering from Sharif University of Technology (SUT), in 2009, 2011, and 2015, respectively. He spent a year as a postdoctoral researcher in Coordinated Science Laboratory at University of Illinois at Urbana-Champaign in 2016-2017. Since 2017, he has been an Assistant Professor of Electrical Engineering at SUT. He is the recipient of INSF starting grant, junior faculty career award, and EPFL IC school fellowship award. His research interests include causal inference, reinforcement learning, and distributed learning.