Workshop on Information Technology
(PRELIMINARY DRAFT)
Scientific description
Over the last century, advances in information technology (IT) have
transformed our daily lives. The desire to share information, to
communicate, has driven unprecedented change in almost every area of
science, engineering and commerce. The most significant developments
in IT are no longer the province of a single adacemic field, but draw
on ideas from diverse sources. Statistics as a scientific discipline,
however, is at the heart of information technology, because it refers
to the collection of products and services that turn data into useful,
meaningful, accessible information.
As part of our
proposed program in statistics, we plan for a workshop covering four
IT problem areas:
| · |
Data Networks
|
|
· |
Databases and Computing
|
| · |
Communications
|
|
· |
Information Systems
|
In each of these areas, researchers are faced with complex systems
built atop a technological and social infrastructure that is
constantly changing. By design, these systems generate vast amounts of
high-dimensional and highly-structured data which are impossible to
describe deterministically. Therefore, the creation of new information
technologies and the characterization of their performance are
fundamentally research problems in large-scale data description and
analysis. The demands of modern IT require sophisticated mathematical
solutions that are able to cope with the richness and variety of human
communications. As examples, consider the new work on characterizing
point processes to model network traffic, the recent advances in
information and coding theory that have extended the capacity of
wireless communications, and the current developments in on-line
learning algorithms that have helped improve chache and prefetching
performance.
The IT component of our proposed MSRI program will bring together
researchers from a variety of disciplines to exchange ideas and forge
new directions for the future.
Outline of topics
-
Data networks
In the last 25 years, we have seen an
explosion in the variety and capabilities of data networks. Wired
networks have achieved a staggering reach, while wireless promises
previously unimaginable mobility and connectivity. Characterizing the
performance of these networks for management and quality of service
assessment is critical. For online monitoring and provisioning,
techniques are being developed for data collection and automated
analysis. Based on these data, researchers are developing strategies
to achieve cost- and service-effective use of network resources. These
require new models for network traffic, and new tools for quickly
identifying bottlenecks. Market-based schemes for network provisioning
like bandwidth and cache trading are also extremely active research
areas. Problems in routing continue to pose challenges for researchers
in optimization. Finally, security issues have been addressed in the
fields of verification, distributed computing, and cryptography, while
new methods for intrustion detection call for sophisticated
statistical models of user behavior. Fields represented include
statistics, queuing and control theory, optimization, economics.
Participants include R. Baranuik, KC Claffy, W. Cleveland, D. Donoho,
N. Duffield, A. Elwalid, S. Floyd, D. Mitra, R. Nowak, V. Paxon,
J. Rexford, Y. Vardi, M. Weinberger, W. Willinger, B. Yu
-
Databases and Computing
Partially due to advances in automated data collection, vast amounts
of scientific and commerce-related data are now housed in databases
around the world. How do we extract useful information from these
repositories? When these databases contain personal data, how do we
guarantee confidentiality and privacy? Algorithms for data analysis
for large-scale problems have been addressed in the literature on data
mining and knowledge discovery, and are receiving more attention from
the general statistics community. The practical infrastructure for
implementing these ideas also poses challenging research questions for
distributed computing, interfaces, inter-system communications and
databases. As the complexity of data stored in these repositories
increases, the demands of statistical models is also greater. McMC?
Quantum?
Participants include D. Temple Lang, J. Friedman, A. Braverman,
T. Hastie, R. Tibshirani, P. Green, J. Rice, Y. Wu, L. Breiman,
D. Pregibon, D. Lambert, G. Wahba, V. Vapnik, M. Wells, J. Hennessey
-
Communication
A crutial piece of the IT infratructure is the communication or
transmission of information between different users of these
technologies. While data Networks address the physical layer of this
link, the application layer includes streaming data analysis, data
compression, image analysis, speech recognition, text-to-speech
translation, and signal processing for wirelss communcations. All
these areas are statistical in nature. While some are old areas
needing novel ideas to meet new constraints (e.g. data compression
and speech recognition), the others are entirely fresh because of the
IT age (e.g. text-to-speech and streaming data analysis). At the
heart of all these fields are the demands for data processing
procedures with IT constraints such as storage space, computation
speed, and process delay, in addition to the traditional statistical
efficiency. Satisfactory solutions in these areas require the combined
efforts of statisticians, engineers, and scientists, which the
workshop promises to bring together.
Participants include J. Kovacevik, S. Savari, A. Ortega, Ramchandran,
S. Verdu, V. Poor, M. Vetterli, M. Jordan, Y. Wu, D. Geman, P. Gong,
D. Nychka, C. Genovese
-
Information systems The connectivity made possible by the
explosion in bandwidth has changed the way we share information. Users
now have access to many different types of data beyond plain
text. Information retrieval, search and recommender systems must
handle the complexity of these new data sources. In addiition, modern
information systems also record a variety of user activities, from
browsing and searching to evaluating and purchasing. Integrating these
data into IR functions can also greatly enhance the effectiveness of
search. Under the broad topic of digital libraries, we see challenging
research areas like adaptive filtering, document clustering, and image
and sound indexing. Also of interest are human-computer infterfaces,
multimedia information processing, Web statistics, and visualization.
Keeping these systems secure is another difficult area requiring
research in storage reliability and virus detection.
Participants include S. Fienberg, D. Madigan, P. Kantor, F. Provost,
J. Lafferty, P. Maes, D. Lewis, H. Varian
Unclaimed names: M. Unser, D. Towsley, Josephine (?, INRA),
S. Kohler, Ji, Lefty, Du Mouchelle, Rissanen
|