BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Ahmet Alacaoglu (UBC) DTSTART:20260303T233000Z DTEND:20260304T003000Z DTSTAMP:20260513T193328Z UID:SFUOR/70 DESCRIPTION:Title: T owards Weaker Variance Assumptions for Stochastic Optimization: A Blast Fr om the Past\nby Ahmet Alacaoglu (UBC) as part of PIMS-CORDS SFU Operat ions Research Seminar\n\nLecture held in ASB 10908.\n\nAbstract\nIn this t alk\, I will present some recent advances for analyzing stochastic optimiz ation methods without the bounded variance assumption. It is well-known th at the bounded variance assumption is violated for even the most standard problems such as linear least squares problem. We will see that the analys is for obtaining optimal rates of convergence under realistic variance ass umptions builds on a connection between the classical literatures for stoc hastic approximation and the Halpern iteration for solving fixed-point pro blems. We will discuss the extensions to proximal algorithms for solving r egularized problems and stochastic convex nonlinear programs\, as well as the required ideas for getting rate guarantees on the last iterate of the algorithm\, which is widely used in practice.\n LOCATION:https://researchseminars.org/talk/SFUOR/70/ END:VEVENT END:VCALENDAR