Dear all,
On Friday April 22 we have Tor Lattimore from DeepMind speaking in the
thematic seminar.
*Tor Lattimore *(DeepMind, http://tor-lattimore.com)
*Friday April 22*, 16h00-17h00
Online on Zoom: *https://uva-live.zoom.us/j/88233925917*
Meeting ID: 882 3392 5917
*Minimax Regret for Partial Monitoring: Infinite Outcomes and
Rustichini's Regret *
The information ratio developed by Russo and Van Roy (2014) is a
powerful tool that was recently used to derive upper bounds on the
regret for challenging sequential decision-making problems. I will talk
about how a generalised version of this machinery can be used to derive
lower bounds and give an application showing that a version of mirror
descent is minimax optimal for partial monitoring using Rustichini's
definition of regret.
Seminar organizers:
Tim van Erven
Botond Szabo
https://mschauer.github.io/StructuresSeminar/
*Upcoming talks:
*Jun. 10,***Julia Olkhovskaya
<https://sites.google.com/view/julia-olkhovskaya/home>*, Vrije Universiteit
--
Tim van Erven<tim(a)timvanerven.nl>
www.timvanerven.nl
Dear all,
This Friday we have Nicolò Cesa-Bianchi from Università degli Studi di
Milano in the thematic seminar.
Note the unusual time, at *11h00* in the morning!
*Nicolò Cesa-Bianchi *(Università degli Studi di Milano,
https://cesa-bianchi.di.unimi.it/)
*Friday April 8*, 11h00-12h00
Online on Zoom: https://uva-live.zoom.us/j/83684658173
Meeting ID: 836 8465 8173
*The power of cooperation in networks of learning agents *
We study the power of cooperation in a network of communicating agents
that solve a learning task. Agents use an underlying communication
network to get information about what the other agents know. In the
talk, we show the extent to which cooperation allows to prove
performance bounds that are strictly better than the known bounds for
non-cooperating agents. Our results are formulated in the setting of
sequential decision-making with partial feedback.
Seminar organizers:
Tim van Erven
Botond Szabo
https://mschauer.github.io/StructuresSeminar/
*Upcoming talks:
*Apr. 22, *Tor Lattimore
<https://mschauer.github.io/StructuresSeminar/#Lattimore>*, DeepMind
Jun. 10,***Julia Olkhovskaya
<https://sites.google.com/view/julia-olkhovskaya/home>*, Vrije Universiteit
--
Tim van Erven<tim(a)timvanerven.nl>
www.timvanerven.nl
Dear Colleagues,
The following event might be appealing our communities.
All the best,
Rui
—
Dept. of Mathematics and Computer Science, TU/e
http://www.win.tue.nl/~rmcastro | +31 40 247 2499
Begin forwarded message:
From: statmathappli <statmathappli(a)inrae.fr<mailto:statmathappli@inrae.fr>>
Subject: StatMathAppli 2022, August 29 till September 2, Fréjus, France : registration is open!
Date: 28 March 2022 at 14:07:03 CEST
To: statmathappli <statmathappli(a)inrae.fr<mailto:statmathappli@inrae.fr>>
Dear colleague,
Please find below the annoucement for the conference StatMathAppli 2022.
Thank you in advance for forwarding it to your lab.
Best regards,
Christophe Giraud and Estelle Kuhn, for the organizing committee
-------------------------------
Dear colleague,
The next edition of StatMathAppli 2022 will take place at Villa Clythia in Fréjus, France, from August 29 till September 2:
statmathappli.mathnum.inrae.fr<https://eur02.safelinks.protection.outlook.com/?url=http%3A%2F%2Fstatmathap…>
The conference provides to European statisticians, including PhD and young researchers, the opportunity to meet, present their work, and initiate collaborations. The program includes (selected) contributed talks, and two courses given by world-renowned researchers.
For the 2022 Edition, we are delighted to welcome as guest lecturer
Rebecca Willett<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwillett.p…> (Chicago University) on "Regularization Induced by Neural Network Architectures" and
Richard Nickl<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.dpmms…> (Cambridge University) on "Bayesian non-linear statistical inverse problems".
Registration is open till Friday, June 17. (https://statmathappli.mathnum.inrae.fr/fr/inscription<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fstatmatha…>)
Please note that, the number of places being limited, the application process consists in a pre-registration on the website, followed by a registration after validation of your application by email from the committee. First arrived, first served.
Abstract submission will be open from Monday, April 25 till Friday, June 24.
Looking forward to welcome you in Fréjus this summer!
Christophe Giraud and Estelle Kuhn, for the organizing committee
Dear all,
This Friday we have Nicolò Cesa-Bianchi from Università degli Studi di
Milano in the thematic seminar.
Note the unusual time, at *11h00* in the morning!
*Nicolò Cesa-Bianchi *(Università degli Studi di Milano,
https://cesa-bianchi.di.unimi.it/)
*Friday April 8*, 11h00-12h00
Online on Zoom: https://uva-live.zoom.us/j/83684658173
Meeting ID: 836 8465 8173
*The power of cooperation in networks of learning agents *
We study the power of cooperation in a network of communicating agents
that solve a learning task. Agents use an underlying communication
network to get information about what the other agents know. In the
talk, we show the extent to which cooperation allows to prove
performance bounds that are strictly better than the known bounds for
non-cooperating agents. Our results are formulated in the setting of
sequential decision-making with partial feedback.
Seminar organizers:
Tim van Erven
Botond Szabo
https://mschauer.github.io/StructuresSeminar/
*Upcoming talks:
*Apr. 22, *Tor Lattimore
<https://mschauer.github.io/StructuresSeminar/#Lattimore>*, DeepMind
Jun. 10,***Julia Olkhovskaya
<https://sites.google.com/view/julia-olkhovskaya/home>*, Vrije Universiteit
--
Tim van Erven<tim(a)timvanerven.nl>
www.timvanerven.nl
Dear all,
This Friday we have Tomer Koren from Tel Aviv University in the
thematic seminar:
*Tomer Koren *(Tel Aviv University, https://tomerkoren.github.io/)
*Friday March 11*, 16h00-17h00
Online on Zoom: https://uva-live.zoom.us/j/85690850169
Meeting ID: 856 9085 0169
Please also join for online drinks after the talk.
*
Benign Underfitting of Stochastic Gradient Descent*
We study to what extent may stochastic gradient descent (SGD) be
understood as a ``conventional'' learning rule that achieves
generalization performance by obtaining a good fit to training data. We
consider the fundamental stochastic convex optimization framework, where
SGD is classically known to minimize the population risk at an optimal
rate, and prove that, surprisingly, there exist problem instances where
the SGD solution exhibits both empirical risk and generalization gap
lower bounded by a universal constant. Consequently, it turns out that
SGD is not algorithmically stable in any sense, and its generalization
ability cannot be explained by uniform convergence or any other
currently known generalization bound technique for that matter (other
than that of its classical analysis). Time permitting, we will discuss
related results for with-replacement SGD, multi-epoch SGD, and
full-batch gradient descent.
Based on joint works with Idan Amir, Roi Livni, Yishay Mansour and Uri
Sherman.
Seminar organizers:
Tim van Erven
Botond Szabo
https://mschauer.github.io/StructuresSeminar/
*Upcoming talks:
*Mar. 25, *Nicolò Cesa-Bianchi
<https://mschauer.github.io/StructuresSeminar/#CesaBianchi>**,
*Università degli Studi di Milano
Apr. 22, *Tor Lattimore
<https://mschauer.github.io/StructuresSeminar/#Lattimore>*, DeepMind
Jun. 10,***Julia Olkhovskaya
<https://sites.google.com/view/julia-olkhovskaya/home>*, Vrije Universiteit
--
Tim van Erven<tim(a)timvanerven.nl>
www.timvanerven.nl
Dear reinforcement learning researcher,
A quick reminder to register for the Belgium-Netherlands workshop on Reinforcement Learning (or, if you have already registered, forward this email to an interested colleague).
We initially aimed for 75 spots, but RL interest in the region has clearly grown, corona regulations have been lifted, and we managed to book a bigger room, so you can still register through: https://rlg.liacs.nl/benerl-2022<https://rlg.liacs.nl/benerl-2022> (currently at 100 registrations).
Hope to see you in June.
Kind regards,
Thomas Moerland & Aske Plaat
Dear all,
We will continue the thematic seminar with a series of online talks. Our
next speaker is Tomer Koren from Tel Aviv University.
*Tomer Koren *(Tel Aviv University, https://tomerkoren.github.io/)
*Friday March 11*, 16h00-17h00
Online on Zoom: https://uva-live.zoom.us/j/85690850169
Meeting ID: 856 9085 0169
Please also join for online drinks after the talk.
**
*
Benign Underfitting of Stochastic Gradient Descent*
We study to what extent may stochastic gradient descent (SGD) be
understood as a ``conventional'' learning rule that achieves
generalization performance by obtaining a good fit to training data. We
consider the fundamental stochastic convex optimization framework, where
SGD is classically known to minimize the population risk at an optimal
rate, and prove that, surprisingly, there exist problem instances where
the SGD solution exhibits both empirical risk and generalization gap
lower bounded by a universal constant. Consequently, it turns out that
SGD is not algorithmically stable in any sense, and its generalization
ability cannot be explained by uniform convergence or any other
currently known generalization bound technique for that matter (other
than that of its classical analysis). Time permitting, we will discuss
related results for with-replacement SGD, multi-epoch SGD, and
full-batch gradient descent.
Based on joint works with Idan Amir, Roi Livni, Yishay Mansour and Uri
Sherman.
Seminar organizers:
Tim van Erven
Botond Szabo
https://mschauer.github.io/StructuresSeminar/
*Upcoming talks:
*Mar. 25, *Nicolò Cesa-Bianchi
<https://mschauer.github.io/StructuresSeminar/#CesaBianchi>**,
*Università degli Studi di Milano
Apr. 22, *Tor Lattimore
<https://mschauer.github.io/StructuresSeminar/#Lattimore>*, DeepMind
Jun. 10,***Julia Olkhovskaya
<https://sites.google.com/view/julia-olkhovskaya/home>*, Vrije Universiteit
**
--
Tim van Erven<tim(a)timvanerven.nl>
www.timvanerven.nl