[Theory] [Talks at TTIC] 11/11 TTIC Colloquium: Tian Li, University of Chicago

[Brandie Jones via Theory]

*When:*        Monday, November 11th at *11:30am CT*


*Where:       *Talk will be given *live, in-person* at

                       TTIC, 6045 S. Kenwood Avenue

                       5th Floor, Room 530


*Virtually:*  via Panopto (livestream
<https://uchicago.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=2a4acdd8-e608-4a39-9f99-b21800f5fe42>
)


*Who: *         Tian Li, University of Chicago

*Title:*          Tilted Empirical Risk Minimization
*Abstract:  *Exponential tilting is a technique commonly used in fields
such as statistics, probability, information theory, and optimization to
shift distributions in a controlled manner. In this talk, I introduce its
usage in machine learning by exploring the use of tilting in risk
minimization. The tilted empirical risk minimization (TERM) framework is a
simple extension of ERM, which uses exponential tilting to flexibly tune
the impact of individual losses. I motivate the tilted objective from
real-world use-cases on addressing representation disparity in federated
learning. I describe several useful theoretical properties of TERM
including its connections with other non-ERM objectives, how it leads to a
tighter Chernoff bound, and a multitude of applications regarding fair and
robust ML. I will conclude the talk by discussing ongoing work on
leveraging this framework to improve sharpness-aware optimization, as well
as other related research.

*Short Bio*: Tian Li is an Assistant Professor at the Computer Science
Department and Data Science Institute at the University of Chicago. Her
research interests are in optimization, trustworthy machine learning, and
efficient learning in large-scale settings. She has spent time at Microsoft
Research Asia, Google Research, and Meta Foundational AI Research Labs. She
was invited to participate in the EECS Rising Stars Workshop, and was
recognized as a Rising Star in Machine Learning/Data Science by multiple
institutions. Her team won the first place in the Privacy-Enhancing
Technologies (PETs) Challenge featured by the White House. She received her
PhD in Computer Science from Carnegie Mellon University and BS degrees in
Computer Science and Economics from Peking University.

*Host: Zhiyuan Li <zhiyuanli at ttic.edu>*

-- 
*Brandie Jones *
*Executive **Administrative Assistant*
Toyota Technological Institute
6045 S. Kenwood Avenue
Chicago, IL  60637
www.ttic.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cs.uchicago.edu/pipermail/theory/attachments/20241104/4f8451cf/attachment.html>