BEGIN:VCALENDAR VERSION:2.0 PRODID:-//wp-events-plugin.com//7.4.0.1//EN TZID:America/New_York X-WR-TIMEZONE:America/New_York BEGIN:VEVENT UID:883@fds.yale.edu DTSTART;TZID=America/New_York:20251029T113000 DTEND;TZID=America/New_York:20251029T130000 DTSTAMP:20251024T134601Z URL:https://fds.yale.edu/events/fds-colloquium-lorenzo-orecchia-chicago/ SUMMARY:FDS Colloquium: Lorenzo Orecchia (Chicago)\, "Variational Character izations of First-Order Algorithms via Self-Duality" DESCRIPTION:\nTalk summary: First-order methods for convex optimization pla y an important role in the efficient deployment of machine learning algori thms. While a large number of such methods exist\, each tuned to the speci fic properties of the problem under consideration\, it is not always clear how to generalize their approach to a new setting or to a different set o f assumptions.\n\n\n\nAt the same time\, certain phenomena\, such accelera tion\, remain unintuitive.\n\n\n\nAn interest approach to systematize the design of first-order method is based on interpreting each method as the d iscretization of a continuous dynamics\, which can be characterized \; variationally. \;This allows us to bring in a large set of tools from the calculus of variation and geometric integration to bear. However\, cur rent interpretations only yield characterizations as stationary points and focus on acceleration\, without explaining other methods such as metric g radient descent or mirror descent. We present improved characterization th at yield characterizations of first-order methods as global minima of natu ral\, duality-gap based\, variational problems. The key idea in our approa ch is self-duality\, which allows us to move beyond Euler-Lagrange equatio ns as a way to introduce variational defined dynamics. We also connect our work with the study of variational formulations for more general dissipat ive systems.\n\n\n\nSpeaker bio: Lorenzo Orecchia is an assistant professo r in the Department of Computer Science at the University of Chicago. Lore nzo’s research focuses on the design of efficient algorithms for fundame ntal computational challenges in machine learning and combinatorial optimi zation. His approach is based on combining ideas from continuous and discr ete optimization into a single framework for algorithm design. Lorenzo obt ained his PhD in computer science at UC Berkeley under the supervision of Satish Rao in 2011\, and was an applied mathematics instructor at MIT unde r the supervision of Jon Kelner until 2014. He was a recipient of the 2014 SODA Best Paper award and a co-organizer of the Simons semester “Bridgi ng Continuous and Discrete Optimization” in Fall 2017.\n\n\n\nWebsite.\n CATEGORIES:Fellows Events,FDS Events,Colloquium LOCATION:Yale Institute for Foundations of Data Science\, Kline Tower 13th Floor\, Room 1327\, New Haven\, CT\, 06511\, United States X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=Kline Tower 13th Floor\, Ro om 1327\, New Haven\, CT\, 06511\, United States;X-APPLE-RADIUS=100;X-TITL E=Yale Institute for Foundations of Data Science:geo:0,0 END:VEVENT BEGIN:VTIMEZONE TZID:America/New_York X-LIC-LOCATION:America/New_York BEGIN:DAYLIGHT DTSTART:20250309T030000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 TZNAME:EDT END:DAYLIGHT END:VTIMEZONE END:VCALENDAR