Principled Tools for Modern Statistical Data Science

Gautam Kamath - Ph.D candidate, Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT

Feb. 5, 2018, 10 a.m. - Feb. 5, 2018, 11 a.m.

McConnell 603


Recent technological developments have given rise to unprecedented quantities and varieties of data. The nature of modern datasets and analyses has given rise to a number of computational and statistical challenges which have not been adequately addressed by classical statistical methods. Desiderata include efficacy in high dimensions, robustness to adversarial attacks or model misspecification, and preserving privacy of the data providers. To address these challenges we must leverage techniques from machine learning, algorithms and statistics. I will discuss principled approaches to address these challenges in the context of fundamental statistical tasks including hypothesis testing and parameter estimation.




Biography: Gautam Kamath is a final-year graduate student in the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT, advised by Constantinos Daskalakis. He received his B.S. in Computer Science from Cornell University in 2012. His research interests are in theoretically-principled methods for statistics, machine learning, and data science, with a focus on settings which arise in modern data analysis. He is recipient of the STOC 2012 Best Student Presentation Award.