BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:Asia/Singapore BEGIN:DAYLIGHT DTSTART:20380119T111407 TZOFFSETFROM:+0800 TZOFFSETTO:+0800 RRULE:FREQ=YEARLY;BYDAY=3TU;BYMONTH=1 TZNAME:+08 END:DAYLIGHT BEGIN:STANDARD DTSTART:19820101T000000 TZOFFSETFROM:+0730 TZOFFSETTO:+0800 RRULE:FREQ=YEARLY;BYDAY=1FR;BYMONTH=1 TZNAME:+08 END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20211112T140748Z UID:AEF35598-367F-46FF-B3EE-AAD992434855 DTSTART;TZID=Singapore:20211112T150000 DTEND;TZID=Singapore:20211112T160000 DESCRIPTION:Research seminar series organized by Sensors Council\, Vehicula r Technology Society\, and Communications Society Singapore Chapter\n\nIn today's networked world\, enormous amounts of data are being collected by edge devices in a distributed fashion. It is often preferred to keep the d ata localized on the edge devices themselves for privacy concerns\, fault- resilience\, or the sheer communication expense of centrally aggregating t he available data. As a consequence\, there has been a rapid emergence of decentralized computational strategies in which global learning tasks are accomplished only through local processing. It this talk we shall review s ome topics in the decentralized optimization setting in which multiple nod es in a network collaborate to minimize the sum of their local loss functi ons.\n\nWe will consider both the centralized learning paradigm (a.k.a. fe derated learning) and the fully distributed paradigm. In both cases\, we w ill introduce the relevant communication constraint for instance (i) a fin ite rate-constraint on the signal transmitted by any node\, (ii) additive noise corrupting the signal received by any node\, and (iii) straggling du e to networking. We will review some of the techniques proposed in the lit erature to address such constraint\, as well as some of my contributions o n this topic. The talk will conclude with some interesting research direct ions at the intersection of distributed optimization and wireless communic ation.\n\nSpeaker(s): Stefano Rini\, \n\nVirtual: https://events.vtools.ie ee.org/m/289263 LOCATION:Virtual: https://events.vtools.ieee.org/m/289263 ORGANIZER:ekurniawan@i2r.a-star.edu.sg SEQUENCE:1 SUMMARY:Research seminar series - Distributed Optimization over Communicati on-constrained Networks: Enabling the Next Generation Machine Intelligence URL;VALUE=URI:https://events.vtools.ieee.org/m/289263 X-ALT-DESC:Description: <br /><p><span style="text-decoration: underline\;" >Research seminar series organized by Sensors Council\, Vehicular Technolo gy Society\, and Communications Society Singapore Chapter</span></p>\n<p>I n today's networked world\, enormous amounts of data are being collected b y edge devices in a distributed fashion. It is often preferred to keep the data localized on the edge devices themselves for privacy concerns\, faul t-resilience\, or the sheer communication expense of centrally aggregating the available data. As a consequence\, there has been a rapid emergence o f decentralized computational strategies in which global learning tasks ar e accomplished only through local processing. It this talk we shall review some topics in the decentralized optimization setting in which multiple n odes in a network collaborate to minimize the sum of their local loss func tions.</p>\n<p>We will consider both the centralized learning paradigm (a. k.a. federated learning) and the fully distributed paradigm. In both cases \, we will introduce the relevant communication constraint for instance (i ) a finite rate-constraint on the signal transmitted by any node\, (ii) ad ditive noise corrupting the signal received by any node\, and (iii) stragg ling due to networking. We will review some of the techniques proposed in the literature to address such constraint\, as well as some of my contribu tions on this topic.&nbsp\; The talk will conclude with some interesting r esearch directions at the intersection of distributed optimization and wir eless communication.</p> END:VEVENT END:VCALENDAR