Montag, 18. November 2013

[HIForum] [Kolloquium] Kolloq: Ralf Herbrich (Amazon Research, Berlin)

Hallo alle,

am Montag in einer Woche gibt es wieder einen Kolloquiums-Vortrag.
Sprecher ist Ralf Herbrich, Director of Machine Learning Science at
Amazon. Er baut gerade Amazons neues Forschungslabor in Berlin auf und
war vorher als leitender Wissenschaftler bei Facebook und bei Microsoft
Research.

Der Vortrag ist auch fuer Mitarbeiter und Studierende interessant, die
mal sehen wollen, was man bei Amazon alles so machen kann. Details siehe
unten.

Viele Gruesse, Ulrike Luxburg




This is an invitation to the next "UHH Informatik Kolloquium"
http://www.informatik.uni-hamburg.de/Info/Kolloquium/index.shtml

SPEAKER: Ralf Herbrich, Ph.D.
Director Machine Learning Science, Amazon Berlin
This talk will be held in English
http://www.herbrich.me/

DATE: MONDAY, 25.11.2013 17:15

PLACE: Konrad-Zuse-Hörsaal, Informatikum, B-201
http://www.informatik.uni-hamburg.de/Info/Campus/index.shtml


-------------------------------------
TOPIC: Large-Scale Machine Learning
-------------------------------------


ABSTRACT:
The last ten years have seen a tremendous growth in Internet-based online
services such as search, advertising, gaming and social networking. Today,
it is important to analyze large collections of user interaction data as a
first step in building predictive models for these services as well as
learn these models in real-time. One of the biggest challenges in this
setting is scale: not only does the sheer scale of data necessitate
parallel processing but it also necessitates distributed models; any
user-specific sets of features in a linear or non-linear model yields
models of a size bigger than can be stored in a single system.

In this talk, I will give an introduction to distributed message passing,
a theoretical framework that can deal both with the distributed inference
and storage of models. After an overview of message passing, I will
discuss and present recent advances in approximate message passing which
allows to control the model size as the amount of training data grows. We
will also review how distributed (approximate) message passing can be
mapped to generalized distributed computing and how modeling constraints
map on the system design. In the second part of the talk, I will give an
overview of the application of these techniques to real-world learning
systems, namely:
1. Gamer ranking and matchmaking in TrueSkill and Halo 3
2. AdPredictor click-through rate learning and prediction in sponsored
search
3. User-action models in Facebook's information distribution and
advertising pipeline



BIO:
Ralf Herbrich is Director of Machine Learning at Amazon Berlin, Germany.
In 2011, he worked at Facebook leading the Unified Ranking and Allocation
team. From 2009 to 2011, he was Director of Microsoft's Future Social
Experiences (FUSE) Lab UK working on the development of computational
intelligence technologies on large online data collections. He holds both
a diploma degree in Computer Science in 1997 and a Ph.D. degree in
Statistics in 2000 from TU Berlin. Ralf's research interests include
Bayesian inference and decision making, computer games, kernel methods and
statistical learning theory. Ralf is one of the inventors of the Drivatars
system in the Forza Motorsport series as well as the TrueSkill ranking and
matchmaking system in Xbox 360 Live. He also co-invented the adPredictor
click-prediction technology launched in 2009 in Bing's online advertising
system.




--

------------------------------------------------------------
Prof. Dr. Ulrike von Luxburg
Department of Computer Science, University of Hamburg

Phone: +49 - (0)40 - 42883 2409
http://www.informatik.uni-hamburg.de/ML/contents/people/luxburg/

------------------------------------------------------------

_______________________________________________
Kolloquium mailing list
Kolloquium@mailhost.informatik.uni-hamburg.de
https://mailhost.informatik.uni-hamburg.de/mailman/listinfo/kolloquium
_______________________________________________
HiForum-Verteiler mailing list
HiForum-Verteiler@mailhost.informatik.uni-hamburg.de
https://mailhost.informatik.uni-hamburg.de/mailman/listinfo/hiforum-verteiler

Donnerstag, 14. November 2013

[HIForum] HIForum News 8/2013

_______________________________________________
HiForum-Verteiler mailing list
HiForum-Verteiler@mailhost.informatik.uni-hamburg.de
https://mailhost.informatik.uni-hamburg.de/mailman/listinfo/hiforum-verteiler

HIForum News 8/2013

Liebe Mitglieder des HIForums,

auch in diesem Monat möchten wir Sie zu  2 Terminen einladen:

Am 16.11. und an den darauffolgenden Donnerstagen findet das IT-Innovationsforum statt.

Das IT-Innovationsforum ist Bestandteil des ITMC-Masterstudiengangs.
Hier sollen die Studierenden eine Übersicht über die aktuellen IT-Innovationen erhalten, ausgewählte Innovationstreiber aus unterschiedlichen Perspektiven kennen- und beurteilen lernen, Aspekte ihrer Entwicklung und Nutzung verstehen und einschätzen sowie Hintergründe für Innovationszyklen und Hypes reflektieren.
 
Die Veranstaltung besitzt Kolloquiumscharakter. Vortragende sind Lehrende des Fachbereichs sowie externe Referenten aus Beratung und Unternehmenspraxis. Die einzelnen Vorträge sind auch für interessierte Externe zugänglich.

Programm 
ITMC
Innovationsforum
 Wintersemester
 2013/14
Donnerstags
 von
 16:15
–
17:45 
in
 Stellingen 
im 
Konrad‐Zuse‐Hörsaal
 in
 Haus
 B
(201)

14.11.    velimo
 ‐
 unsere
 Entertainment‐Lösung
 für
 alle
 Mobile‐Plattformen


              Patrick
 Götze,
 Lufthansa
 Systems

21.11.    Anforderungen
 an
 Rechenzentren 
in 
modernen
 Netzarchitekturen

              Philipp
 Frenzel
 und
 Dirk
 Krone, 
Dataport

28.11.    Lessons
 learned 
‐
 modellbasierte
 Entwicklung
 in
 der
 Praxis


              Niels 
Stargardt
 und 
Stefan 
Vocke,
 PPI


die weiteren Termine dieser Veranstaltungsreihe finden Sie hier:
http://agis-www.informatik.uni-hamburg.de/fileadmin/itmc/pdf/Info/Prog_IT-Innovationsforum__WS_13_14.pdf

am Montag, 25.11.2013 findet das nächste Informatik Kolloquium statt:

Ralf Herbrich Ph. D. (Director of Machine Learning Science at Amazon): Large-Scale Machine Learning
 http://www.informatik.uni-hamburg.de/Info/Kolloquium/ws13/herbrich.shtml

Hier noch einmal die Einladung zur Carl Friedrich von Weizsäcker-Friedensvorlesung:
Die Ringvorlesung, für die wir namhafte Referenten aus Wissenschaft, Wirtschaft und Politik gewinnen konnten, wird von Fachbereich Informatik, dem Carl Friedrich von Weizsäcker-Zentrum für Naturwissenschaft und Friedensforschung (ZNF) und dem Institut für Friedensforschung und Sicherheitspolitik (IFSH), gemeinsam veranstaltet.
Die Fachvorträge der Ringvorlesung finden jeweils mittwochs um 16 Uhr im Hauptgebäude der Universität, Agathe-Lasch-Hörsaal (ESA-B), Edmund-Siemers-Allee 1, 20146 Hamburg, statt.
Weitere Informationen finden Sie unter http://www.informatik.uni-hamburg.de/friedensvorlesung/.

Die nächsten beiden Termine:
20.11.2013     Schutz Kritischer Infrastrukturen als Element der Cybersicherheit
Christoph Unger, Präsident des Bundesamtes für Bevölkerungsschutz und Katastrophenhilfe
27.11.2013     Die Cyber-Sicherheitsarchitektur für Deutschland
Dr. Günther Welsch, Fachbereichsleiter Koordination und Steuerung im Bundesamt für Sicherheit in der Informationstechnik (BSI)


Mit besten Grüßen
Michael Schudy, Angela Schwabl, Dirk Martinssen

HIForum - Hamburger Informatik-Forum e.V.
Vorstand: Dipl.-Inform. Michael Schudy (Vorsitzender), 
Dipl.-Inform. Angela Schwabl,
Dipl.-Inform. Dirk Martinssen
Vereinsregister Hamburg VR 16010
c/o Fachbereich Informatik der Universität Hamburg
Vogt-Kölln-Straße 30
22527 Hamburg

Montag, 11. November 2013

[HIForum] [Kolloquium] UHH Inform. Kolloq., Mo, 25 November 2013, Ralf Herbrich, Ph.D., Amazon Berlin

This is an invitation to the next "UHH Informatik Kolloquium"
http://www.informatik.uni-hamburg.de/Info/Kolloquium/index.shtml


SPEAKER: Ralf Herbrich, Ph.D.
Director Machine Learning Science, Amazon Berlin
This talk will be held in English
http://www.herbrich.me/

DATE: MONDAY, 25.11.2013 17:15

PLACE: Konrad-Zuse-Hörsaal, Informatikum, B-201
http://www.informatik.uni-hamburg.de/Info/Campus/index.shtml


-------------------------------------
TOPIC: Large-Scale Machine Learning
-------------------------------------


ABSTRACT:
The last ten years have seen a tremendous growth in Internet-based online
services such as search, advertising, gaming and social networking. Today,
it is important to analyze large collections of user interaction data as a
first step in building predictive models for these services as well as
learn these models in real-time. One of the biggest challenges in this
setting is scale: not only does the sheer scale of data necessitate
parallel processing but it also necessitates distributed models; any
user-specific sets of features in a linear or non-linear model yields
models of a size bigger than can be stored in a single system.

In this talk, I will give an introduction to distributed message passing,
a theoretical framework that can deal both with the distributed inference
and storage of models. After an overview of message passing, I will
discuss and present recent advances in approximate message passing which
allows to control the model size as the amount of training data grows. We
will also review how distributed (approximate) message passing can be
mapped to generalized distributed computing and how modeling constraints
map on the system design. In the second part of the talk, I will give an
overview of the application of these techniques to real-world learning
systems, namely:
1. Gamer ranking and matchmaking in TrueSkill and Halo 3
2. AdPredictor click-through rate learning and prediction in sponsored
search
3. User-action models in Facebook's information distribution and
advertising pipeline



BIO:
Ralf Herbrich is Director of Machine Learning at Amazon Berlin, Germany.
In 2011, he worked at Facebook leading the Unified Ranking and Allocation
team. From 2009 to 2011, he was Director of Microsoft's Future Social
Experiences (FUSE) Lab UK working on the development of computational
intelligence technologies on large online data collections. He holds both
a diploma degree in Computer Science in 1997 and a Ph.D. degree in
Statistics in 2000 from TU Berlin. Ralf's research interests include
Bayesian inference and decision making, computer games, kernel methods and
statistical learning theory. Ralf is one of the inventors of the Drivatars
system in the Forza Motorsport series as well as the TrueSkill ranking and
matchmaking system in Xbox 360 Live. He also co-invented the adPredictor
click-prediction technology launched in 2009 in Bing's online advertising
system.



CONTACT:
Prof. Dr. Ulrike von Luxburg
Department of Computer Science, University of Hamburg
Phone: +49-(0)40-42883-2409
http://www.informatik.uni-hamburg.de/ML/contents/people/luxburg/


_______________________________________________
Kolloquium mailing list
Kolloquium@mailhost.informatik.uni-hamburg.de
https://mailhost.informatik.uni-hamburg.de/mailman/listinfo/kolloquium
_______________________________________________
HiForum-Verteiler mailing list
HiForum-Verteiler@mailhost.informatik.uni-hamburg.de
https://mailhost.informatik.uni-hamburg.de/mailman/listinfo/hiforum-verteiler