BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:Ahmed Khaled (Cairo University)
DTSTART:20200513T130000Z
DTEND:20200513T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/1
DESCRIPTION:Title: On
the Convergence of Local SGD on Identical and Heterogeneous Data\nby A
hmed Khaled (Cairo University) as part of Federated Learning One World Sem
inar\n\nAbstract: TBA\n
LOCATION:https://researchseminars.org/talk/FLOW/1/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Blake Woodworth (TTIC)
DTSTART:20200520T130000Z
DTEND:20200520T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/2
DESCRIPTION:Title: Is
local SGD better than minibatch SGD?\nby Blake Woodworth (TTIC) as par
t of Federated Learning One World Seminar\n\nAbstract: TBA\n
LOCATION:https://researchseminars.org/talk/FLOW/2/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Dimitris Papailiopoulos (University of Wisconsin-Madison)
DTSTART:20200527T130000Z
DTEND:20200527T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/3
DESCRIPTION:Title: Rob
ustness in federated learning may be impossible without an all-knowing cen
tral authority\nby Dimitris Papailiopoulos (University of Wisconsin-Ma
dison) as part of Federated Learning One World Seminar\n\nAbstract: TBA\n
LOCATION:https://researchseminars.org/talk/FLOW/3/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Sai Praneeth Karimireddy (EPFL)
DTSTART:20200610T130000Z
DTEND:20200610T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/4
DESCRIPTION:Title: Sto
chastic controlled averaging for federated learning\nby Sai Praneeth K
arimireddy (EPFL) as part of Federated Learning One World Seminar\n\nAbstr
act: TBA\n
LOCATION:https://researchseminars.org/talk/FLOW/4/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Filip Hanzely (KAUST)
DTSTART:20200617T130000Z
DTEND:20200617T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/5
DESCRIPTION:Title: Fed
erated learning of a mixture of global and local models: Local SGD and opt
imal algorithms\nby Filip Hanzely (KAUST) as part of Federated Learnin
g One World Seminar\n\nAbstract: TBA\n
LOCATION:https://researchseminars.org/talk/FLOW/5/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Hadrien Hendrikx (École Normale Supérieure & INRIA)
DTSTART:20200624T130000Z
DTEND:20200624T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/6
DESCRIPTION:Title: Sta
tistical preconditioning for federated learning\nby Hadrien Hendrikx (
École Normale Supérieure & INRIA) as part of Federated Learning One Worl
d Seminar\n\nAbstract: TBA\n
LOCATION:https://researchseminars.org/talk/FLOW/6/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Alireza Fallah (MIT)
DTSTART:20200701T130000Z
DTEND:20200701T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/7
DESCRIPTION:by Alireza Fallah (MIT) as part of Federated Learning One Worl
d Seminar\n\nAbstract: TBA\n
LOCATION:https://researchseminars.org/talk/FLOW/7/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Jakub Konečný (Google)
DTSTART:20200708T130000Z
DTEND:20200708T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/8
DESCRIPTION:Title: On
the outsized importance of learning rates in local update methods\nby
Jakub Konečný (Google) as part of Federated Learning One World Seminar\n
\n\nAbstract\nIn this work\, we study a family of algorithms\, which we re
fer to as local update methods\, that generalize many federated learning a
nd meta-learning algorithms. We prove that for quadratic objectives\, loca
l update methods perform stochastic gradient descent on a surrogate loss f
unction which we exactly characterize. We show that the choice of client l
earning rate controls the condition number of that surrogate loss\, as wel
l as the distance between the minimizers of the surrogate and true loss fu
nctions. We use this theory to derive novel convergence rates for federate
d averaging that showcase this trade-off between the condition number of t
he surrogate loss and its alignment with the true loss function. We valida
te our results empirically\, showing that in communication-limited setting
s\, proper learning rate tuning is often sufficient to reach near-optimal
behavior. We also present a practical method for automatic learning rate d
ecay in local update methods that helps reduce the need for learning rate
tuning\, and highlight its empirical performance on a variety of tasks and
datasets.\n
LOCATION:https://researchseminars.org/talk/FLOW/8/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Sashank Reddi (Google)
DTSTART:20200715T130000Z
DTEND:20200715T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/9
DESCRIPTION:Title: Ada
ptive federated optimization\nby Sashank Reddi (Google) as part of Fed
erated Learning One World Seminar\n\n\nAbstract\nFederated learning is a d
istributed machine learning paradigm in which a large number of clients co
ordinate with a central server to learn a model without sharing their own
training data. Due to the heterogeneity of the client datasets\, standard
federated optimization methods such as Federated Averaging (FedAvg) are of
ten difficult to tune and exhibit unfavorable convergence behavior. In non
-federated settings\, adaptive optimization methods have had notable succe
ss in combating such issues. In this work\, we propose federated versions
of adaptive optimizers\, including Adagrad\, Adam\, and Yogi\, and analyze
their convergence in the presence of heterogeneous data for general nonco
nvex settings. Our results highlight the interplay between client heteroge
neity and communication efficiency. We also perform extensive experiments
on these methods and show that the use of adaptive optimizers can signific
antly improve the performance of federated learning.\n
LOCATION:https://researchseminars.org/talk/FLOW/9/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Nati Srebro (TTIC)
DTSTART:20200722T130000Z
DTEND:20200722T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/10
DESCRIPTION:Title: He
terogeneity and pluralism in distributed learning\nby Nati Srebro (TTI
C) as part of Federated Learning One World Seminar\n\nAbstract: TBA\n
LOCATION:https://researchseminars.org/talk/FLOW/10/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Krishna Pillutla (University of Washington)
DTSTART:20200729T130000Z
DTEND:20200729T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/11
DESCRIPTION:Title: Ro
bust Aggregation for Federated Learning\nby Krishna Pillutla (Universi
ty of Washington) as part of Federated Learning One World Seminar\n\n\nAbs
tract\nKrishna Pillutla\, Sham M. Kakade\, Zaid Harchaoui. Robust Aggregat
ion for Federated Learning\, arXiv:1912.13445\, 2019.\n\nYassine Laguel\,
Krishna Pillutla\, Jérôme Malick\, Zaid Harchaoui. Device Heterogeneity
in Federated Learning: A Superquantile Approach\, arXiv:2002.11223\, 2020.
\n
LOCATION:https://researchseminars.org/talk/FLOW/11/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Konstantin Mishchenko (KAUST)
DTSTART:20200805T130000Z
DTEND:20200805T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/12
DESCRIPTION:Title: Lo
cal decentralized gradient descent with fast convergence\nby Konstanti
n Mishchenko (KAUST) as part of Federated Learning One World Seminar\n\nAb
stract: TBA\n
LOCATION:https://researchseminars.org/talk/FLOW/12/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Peter Kairouz (Google)
DTSTART:20200812T130000Z
DTEND:20200812T140000Z
DTSTAMP:20260422T225752Z
UID:FLOW/13
DESCRIPTION:Title: Fe
derated analytics\nby Peter Kairouz (Google) as part of Federated Lear
ning One World Seminar\n\nAbstract: TBA\n
LOCATION:https://researchseminars.org/talk/FLOW/13/
END:VEVENT
END:VCALENDAR