Events - 2008
Departmental
Colloquium
Access Control Polynomial-based Security Support for
Trusted Collaborative Computing (TCC)
Dr. Xukai Zou
Department of Computer and Information Science
Indiana University-Purdue University
Indianapolis
Information and communication technologies along with
society’s drive for collaboration in the modern world make “collaborative
computing” and its applications possible and even necessary. Trust in this
environment will eventually determine its success and popularity due to
human and corporate desire for privacy and integrity. The current
Internet is by design not security-oriented. Security patches and
enhancement mechanisms result in more security vulnerabilities. Compared
to the two-party interaction model (i.e., the client-server service
model), collaborative computing environments are group-oriented, involve a
large number of users and shared resources, and are complex, dynamic,
distributed, and heterogeneous, including even hostile elements. Systems
experience failures due to faults and attacks from hostile entities. More
seriously, there are dangerous attacks from malicious internal members.
Consequently, building a trusted collaborative computing (TCC) environment
is extremely difficult and requires a long-term persevering endeavor. TCC
is a new research and application paradigm that features group-oriented
communication and resource sharing (in a controlled manner), applies to a
broad range of applications such as multi-party military actions, tele-conferencing,
tele-medicine, interactive and collaborative decision making,
grid-computing, information distribution, and pay-per-view audio/video
service, and is facing many open and challenging problems. In this talk,
we will discuss some fundamental security functions involved in TCC
including secure group communication (SGC) and controlled access to shared
resources and their state-of-the-art key management solutions. We will
also present a new cryptographic primitive, Access Control Polynomial,
which can support various security functions in TCC uniformly and
illustrate its seamless integration with the Veterans Affairs’ medical
information system.
About the Speaker: Dr. Xukai Zou is an assistant
professor in the Department of Computer and Information Science, Indiana
University-Purdue University Indianapolis. His current research includes
applied cryptography and network security, trusted collaborative
computing, in particular, secure group communication (SGC) and access
control in distributed and heterogeneous environments, and security in
wireless networks. He has published two books and about 30 articles
related to these topics. Recently, he designed an efficient and practical
group key management which uniformly supports all four fundamental TCC
security functions. Dr. Zou is a recipient of an NSF cyber trust grant for
his project "Secure Group Communication over Wired/wireless Networks." He
is also an editor, a program committee chair, a technical committee member,
and a reviewer for a number of international journals and conferences.
Tuesday, May 13, 2008
1:00 p.m.–2:00 p.m.
Department Conference Room
Departmental
Colloquium
Machine Learning for Autonomic
Management in Networked Computing Systems
Dr. Cheng-Zhong Xu
Department of Electrical and Computer Engineering
Wayne State University
Computer systems keep growing in scale and complexity.
It becomes a big challenge to get a good understanding of large-scale
systems' dynamic behavior via the traditional divide-and-conquer
analytical approach. Optimizing one component may compromise the others,
leading to overall performance degradation.
In this talk, I will present our recent studies on the
use of machine learning technologies to tackle performance and reliability
problems in networked computer systems. The first part will be on a
stochastic model to quantify failure event correlation in both space and
time domains for proactive failure management. The second part will be on
a Bayesian network model to correlate OS- and hardware-level parameters to
application-level performance metrics for system anomaly detection and
location. Empirical models built using statistical learning have great
potential to help overcome the challenges of scale and complexity in
current and future computer systems.
About the Speaker: Dr. Cheng-Zhong Xu is a
Professor in the Department of Electrical and Computer Engineering of
Wayne State University and the Director of the Laboratory for Networked
Computing Systems. Dr. Xu's research interests include resource management
in distributed and parallel systems, high-performance cluster computing,
and scalable and secure Internet services. He has published more than 120
peer-reviewed journal and conference papers in these areas. He is the lead
co-author of the book Load Balancing in Parallel Computers (Kluwer
Academic, 1996). It was one of the first research monographs that
addressed the load-balancing issue systematically. Dr. Xu’s most recent
book Scalable and Secure Internet Services and Architecture
(Chapman & Hall/CRC Press, 2005) provides an in-depth analysis of Internet
services in a unified framework from the performance perspective. Dr. Xu
is currently serving in five editorial boards, including IEEE
Transactions on Parallel and Distributed Systems and Journal of
Parallel and Distributed Computing. He is a recipient of the Faculty
Research Award of Wayne State University in 2000, the President's Award
for Excellence in Teaching in 2002, and the Career Development Chair Award
in 2003. Dr. Xu received a Ph.D.in Computer Science from the University of
Hong Kong in 1993.
Friday, March 14, 2008
1:00 p.m.–2:00 p.m.
Department Conference Room
Departmental
Colloquium
Mathematical and Computational Aspects
of "Sequencing" Proteins by Mass-Spectrometry
Dr. Andrey Gorin
Senior Research Staff Member
Computer Science and Mathematics Division
Oak Ridge National Laboratory
Mass-spectrometry (MS) is a powerful experimental
technology for "sequencing" proteins in complex biological mixtures.
Computational methods are essential for interpretation of MS data, and a
number of hard theoretical questions remain weakly understood due to the
intrinsic complexity of the related algorithms. The talk describes a
mathematical and computational framework for an analytical measure of
peptide identification reliability in so-called database search methods.
The approach is based on database searches with mass tags as sequences of
masses (m1 m2 ... mn),
where individual values are
distances between spectral lines of identical chemical nature. First, we
define the p-function as the probability of finding a random match
between any given tag and a protein database and confirm its quality with
extensive tag search experiments. We then discuss properties of the p-function
and its implications for finding highly reliable IDs in MS experiments.
Finally we outline a possibility to evaluate analytically Sequest
X-correlation (the holy grail of the currently popular algorithms) and
describe some of the remaining problems in the area of computational MS
proteomics.
Thursday, February 28, 2008
3:15 p.m.–5:00 p.m.
Department Conference Room
Molecular Basis of
Disease Distinguished Lecture Series
Finding the Right Genes
Dr. Lawrence O. Hall
Professor
Department of Computer Science and Engineering
University of South Florida
In pattern recognition problems, there are typically
more examples or patterns than there are features. Even so, feature
selection is typically done to reduce the number of features. With
micro-array data there are typically many more genes or features than
there are tissue samples or example patterns. Ratios of 10 or 100 to 1 are
not uncommon. In order to build predictive models, it is necessary to find
the best or right genes for the purpose. Of course, the most predictive
genes will be of medical and biological significance also. This talk
explores approaches to selecting genes to be able to predict a disease or
a likely prognosis, including effective pre-processing. A method of
feature perturbation which can be applied in conjunction with any learning
algorithm is compared with other approaches such as recursive feature
elimination on a publicly available data set.
Thursday, January 17, 2008
10:00 a.m.–11:00 a.m.
Commerce Club, 14th floor, Sterne Room
