[Theory] UPDATED LINK: Young Researcher Series: 5/18 Annie Marsden, Stanford University

[Brandie Jones]

*When:    *  Wednesday, May 18th at *10:30am CT*



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

                   TTIC, 6045 S. Kenwood Avenue

                     5th Floor, Room 530


*Where:     *via Panopto (Livestream)
<https://uchicago.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=5146d512-bef7-43ec-b3d0-ae9900ff1f41>
 - *UPDATED*



*Who:         *Annie Marsden, Stanford University

*Title:         *Efficient Convex Optimization Requires Superlinear Memory

*Abstract:*  How much memory is necessary to efficiently optimize convex
functions? In this talk I will present a result which establishes that
superlinear memory is required to be competitive with the best known
quadratic memory algorithms. Specifically, any first-order algorithm must
use either O(d^{1.25 – delta}) memory or make at least Omega(d^{1 + (4/3)
delta}) queries. This resolves (a part of) a COLT 2019 open problem of
Woodworth and Srebro.

This is joint work with Vatsal Sharan, Aaron Sidford, and Gregory Valiant.

Bio:       Annie is a fifth year Ph.D. student in Computer Science at
Stanford University co-advised by Gregory Valiant and John Duchi. Her
interests are broadly in the intersection of algorithms, statistics,
optimization, and machine learning. Before Stanford, she was fortunate to
work with John Lafferty at the University of Chicago. Prior to that,
she received an MPhil in Scientific Computing at the University of
Cambridge on a Churchill Scholarship where she was advised by Sergio
Bacallado. She received a B.S. in Mathematics and B.A. in Chemistry at the
University of Chicago where she was lucky to enjoy the mentorship of
Stephen Berry, László Babai, and Robert Soare.

Host: Nathan Srebro <nati at ttic.edu>
*Brandie Jones *
*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/20220518/a7c6071f/attachment-0001.html>