|
|
 |
|
|
| The Next Generation of Neural Networks |
 |
Google Tech Talks
November, 29 2007
In the 1980's, new learning algorithms for
neural networks promised to
solve difficult classification tasks, like
speech or object recognition,
by learning many layers of non-linear
features. The results were
disappointing for two reasons: There was
never enough labeled data to
learn millions of complicated features and
the learning was much too slow
in deep neural networks with many layers of
features. These problems can
now be overcome by learning one layer of
features at a time and by
changing the goal of learning. Instead of
trying to predict the labels,
the learning algorithm tries to create a
generative model that produces
data which looks just like the unlabeled
training data. These new neural
networks outperform other machine learning
methods when labeled data is
scarce but unlabeled data is plentiful. An
application to very fast
document retrieval will be described.
Speaker: Geoffrey Hinton
Geoffrey Hinton received his BA in
experimental psychology from Cambridge in
1970 and his PhD in Artificial Intelligence
from Edinburgh in 1978. He did
postdoctoral work at Sussex University and
the University of California San
Diego and spent five years as a faculty
member in the Computer Science
department at Carnegie-Mellon University. He
then became a fellow of the
Canadian Institute for Advanced Research and
moved to the Department of
Computer Science at the University of
Toronto. He spent three years from 1998
until 2001 setting up the Gatsby
Computational Neuroscience Unit at University
College London and then returned to the
University of Toronto where he is a
University Professor. He holds a Canada
Research Chair in Machine Learning. He
is the director of the program on "Neural
Computation and Adaptive Perception"
which is funded by the Canadian Institute for
Advanced Research.
Geoffrey Hinton is a fellow of the Royal
Society, the Royal Society of Canada,
and the Association for the Advancement of
Artificial Intelligence. He is an
honorary foreign member of the American
Academy of Arts and Sciences, and a
former president of the Cognitive Science
Society. He received an honorary
doctorate from the University of Edinburgh in
2001. He was awarded the first
David E. Rumelhart prize (2001), the IJCAI
award for research excellence
(2005), the IEEE Neural Network Pioneer award
(1998) and the ITAC/NSERC award
for contributions to information technology
(1992).
A simple introduction to Geoffrey Hinton's
research can be found in his
articles in Scientific American in September
1992 and October 1993. He
investigates ways of using neural networks
for learning, memory, perception and
symbol processing and has over 200
publications in these areas. He was one of
the researchers who introduced the
back-propagation algorithm that has been
widely used for practical applications. His
other contributions to neural
network research include Boltzmann machines,
distributed representations,
time-delay neural nets, mixtures of experts,
Helmholtz machines and products of
experts. His current main interest is in
unsupervised learning procedures
for neural networks with rich sensory input.
Tags : google techtalks techtalk engedu talk talks googletechtalks education |
|
Affichage : 41784
Durée : 3563 s |
| Next Generation All-IP Telecom Networks: Quality of Service Challenges and Is... |
 |
Google Tech Talks
January, 14 2008
ABSTRACT
The SIP-based IP Multimedia Subsystem (IMS),
while recently introduced, has become one of
the primary distinguishing features of the
next generation of mobile telecommunication
systems. IMS allows mobile operators to offer
advanced value-added services - like VoIP,
so-called push-to-media, video, interactive
gaming, and mobile banking - to their
customers timely and efficiently. Google's
plans to enter the wireless world open up a
world of possibilities for offering customers
and businesses advanced services such as
targeted location-based services and
advertisements through the IMS framework.
Deploying IMS, however, is a non-trivial
task. The core challenge for the telecom
industry has been and will be the integration
of the current radio access network (RAN) and
IP transport infrastructure with the IMS
domain. Within standardization bodies,
efforts are underway to address the issues
for call setup and mobility signaling, while
developing unified user profile management
and Quality of Service (QoS) architectures.
The real goal is a standardized, IMS-centric,
end-to-end unified signaling architecture.
To this end, this presentation provides an
overview of IMS and QoS signaling over
integrated RAN and IMS domains. By using an
exemplary family media service, aspects and
specifics of the end-to-end QoS invocation,
control and policy enforcement, including
roaming scenarios, are demonstrated. Based on
laboratory measurements performed at
Sprint-Nextel aided with simulations, the
Post Dial Delay (PDD) delay is evaluated and
some practical recommendations for delay
reduction are presented. The presentation
will conclude with discussion of open issues
and viable solutions. This presentation
should be of interest to Googlers who work on
mobile related projects and intend to have a
big picture of next generation mobile systems
such as application development, and service
and system integration with wireless
operators.
This presentation is based on the article S.
Zaghloul, A. Jukan, W. Alanqar:
"Extending QoS from Radio Access to
all-IP Core in 3G Networks - An Operator's
Perspective," IEEE Communications
Magazine, Sept 2007.
Speaker: Said Zaghloul
Fulbright alumnus and former
Telecommunication Design Engineer at
Sprint-Nextel
Research Staff Member, PhD Candidate
Institute of Computer and Communication
Network Engineering
Technical University Carolo-Wilhelmina of
Braunschweig, Germany
Tags : google techtalks techtalk engedu talk talks googletechtalks education |
|
Affichage : 14368
Durée : 3390 s |
| Social networks and trust : NetTrust |
 |
Google Tech Talks
February, 28 2008
ABSTRACT
NetTrust is a system that embeds social
context in browsing by combining individual
histories, social networks, and explicit
ratings. NetTrust combines an implicit and
explicit means of data collection. This trust
based system uses shared browsing histories
from a user's self-selected social networks
to create both explicit and implicit data
collection. NetTrust targets the human
element of trust. It projects how a social
network can signal meaningful trust
information that can make an educative
browsing experience. NetTrust allows an
individual to select their own trusted
sources of information and rate particular
sites as trustworthy (or not). NetTrust
allows an individual to select their own
trusted authoritative sources of information
from a market of ratings agencies and combine
these ratings with the reputation information
from their individual social network. This
paper will present the Net Trust system; the
dorm-based homophily tests with implications
and the undergraduate-focused user testing.
Speaker: Professor L. Jean Camp
Professor L. Jean Camp is the author of Trust
and Risk in Internet Commerce (MIT Press),
Economics of Identity Theft (Springer) and
the editor of Economics of Information
Security (Kluwer Academic). She has authored
over one hundred works, including seventy
peer-reviewed works and eighteen book
chapters. In addition to presentations at
peer-reviewed venues, she has made scores of
invited presentations on four continents. Her
service has included the Board of Directors
of Computer Professionals for Social
Responsibility, the Board of Governors of the
IEEE Society on Social Implications of
Technology, Senior Member of the IEEE, and
longstanding member of the USACM. See
http://www.ljean.com/cv.html for more
detailed information and full text of various
publications.
Tags : google techtalks techtalk engedu talk talks googletechtalks education |
|
Affichage : 5151
Durée : 3440 s |
|
|
|
|
|
|
|
|
|
|
 |
| |
|
