BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Department of Applied Mathematics and Statistics - ECPv6.5.1.6//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:Department of Applied Mathematics and Statistics
X-ORIGINAL-URL:https://engineering.jhu.edu/ams
X-WR-CALDESC:Events for Department of Applied Mathematics and Statistics
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20230312T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20231105T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230309T133000
DTEND;TZID=America/New_York:20230309T143000
DTSTAMP:20240715T055746
CREATED:20230306T182646Z
LAST-MODIFIED:20230306T182646Z
UID:44869-1678368600-1678372200@engineering.jhu.edu
SUMMARY:AMS Weekly Seminar | Assistant Professor Ben Grimmer
DESCRIPTION:Location: Gilman 132 \n\n\nWhen: March 9th at 1:30 p.m. \n\n\nTitle: Scalable\, Projection-Free Optimization Methods \n\n\nAbstract: We will introduce an approach to constrained optimization replacing orthogonal projections with much cheaper radial ones. This results in new first-order methods that are (i) scalable\, sporting minimal per iteration costs\, (ii) always yield fully feasible solutions\, and (iii) applicable to a wide family of potentially nonconvex constraints. For example\, when applied to linear programs\, iteration costs amount to one matrix multiplication instead of more costly pivoting or matrix inverses. For semidefinite programming\, iterations require computing one eigenvector\, avoiding full decompositions. Beyond these classic settings\, modern applications with nonconvex objectives and constraints will be discussed. The key mathematical insight underlying these methods is a new duality relating constrained optimization problems to unconstrained “radially dual” problems. From this\, we will present accelerated projection-free first-order methods using smoothness\, strong convexity\, or other common structures in the primal objective or constraints.
URL:https://engineering.jhu.edu/ams/event/ams-weekly-seminar-assistant-professor-ben-grimmer/
CATEGORIES:Seminars and Lectures
END:VEVENT
END:VCALENDAR