BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Date iCal//NONSGML kigkonsult.se iCalcreator 2.20.2//
METHOD:PUBLISH
X-WR-CALNAME;VALUE=TEXT:Eventi DIAG
BEGIN:VTIMEZONE
TZID:Europe/Paris
BEGIN:STANDARD
DTSTART:20211031T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20210328T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.23406.field_data.0@dis.uniroma1.it
DTSTAMP:20260306T112551Z
CREATED:20210621T140316Z
DESCRIPTION:Abstract:We present a novel view on principal component analysi
 s (PCA)\, equivalently singular value decomposition (SVD)\, as a competiti
 ve game in which each approximate singular vector is controlled by a playe
 r whose goal is to maximize their own utility function. We analyze the pro
 perties of this EigenGame and the behavior of its gradient based updates. 
 The resulting algorithm -- which combines elements from Oja's rule with a 
 generalized Gram-Schmidt orthogonalization -- is naturally decentralized a
 nd hence parallelizable through message passing. We demonstrate the scalab
 ility of the algorithm by conducting principal component analyses of large
  image datasets and neural network activations. We discuss how this new vi
 ew of SVD as a differentiable game can lead to further algorithmic develop
 ments and insights.This talk is based on joint work with Brian McWilliams\
 , Claire Vernade\, and Thore Graepel -- https://arxiv.org/abs/2010.00554 (
 EigenGame - ICLR ‘21). The paper received an 'Outstanding Paper Award' at 
 ICLR'21.Deepmind's Blog post: https://deepmind.com/blog/article/EigenGameS
 hort Bio:Ian is a Research Scientist on the Multiagent team at DeepMind. H
 is research focuses primarily on two questions. How should agents behave i
 n a group\, be it a competitive\, mixed-motive\, or cooperative setting? A
 nd should individual agents themselves (including their constituent tools 
 and algorithms) be considered multi-agent systems in their own right? He s
 tudied mechanical engineering and applied math (BS/MS) at Northwestern Uni
 versity (2011) and obtained his MS/PhD in computer science from the Univer
 sity of Massachusetts at Amherst (2018).
DTSTART;TZID=Europe/Paris:20210713T140000
DTEND;TZID=Europe/Paris:20210713T153000
LAST-MODIFIED:20210621T150601Z
LOCATION:https://uniroma1.zoom.us/j/84848759900?pwd=a3NwWEFDZTlpZHhWbkRTQmF
 mdkowUT09
SUMMARY:EigenGame: PCA as a Nash Equilibrium - Ian Gemp (DeepMind)
URL;TYPE=URI:https://dis.uniroma1.it/node/23406
END:VEVENT
END:VCALENDAR