The Teacher Research Group, headed by Professor Benjamin W. Wah, works on the theory,
design, analysis, evaluation, implementation, and application of nonlinear optimization algorithms.
The applications of the results include
nonlinear programming,
temporal planning,
digital signal processing,
artificial neural networks,
operations research,
evolutionary computations,
load balancing,
parallel-architecture design,
computer network-protocol design,
computer-aided design,
circuit testing, and
computer vision.
The current and past research is classified into six areas listed on the left.
To see a summary, click on one of the areas to see a summary. Each summary will lead to various lists of
publications, some of which are in PostScript and/or PDF that can be downloaded by FTP.
The current research interests include the following:
- Nonlinear programming, mixed-integer programming, and applications in financial engineering.
- Multimedia signal processing, voice-over-IP, and computer networks.
- Artificial Intelligence, planning and scheduling, and satisfiability.
The following is a brief summary of the current research:
- The first project focuses on the study of the theory and algorithms for solving nonlinear constrained optimization
problems in discrete, continuous, or mixed spaces. Based on a new necessary and sufficient condition for characterizing
constrained local optima and by exploiting the localities of constraints in these problems, his team has developed
new algorithms for solving large-scale application problems in temporal planning, satisfiability, and nonlinear
programming that are too large for other solvers.
See a story on our current work.
-
The second project aims at developing new protocols and coding methods for concealing losses that occur during real-time
transmissions of multimedia data (voice, speech, images, and video) over unreliable IP networks, such as the Internet
and wireless networks. Since video and audio transmissions may tolerate some degrees of loss, his team focuses on
studying trade-offs among the design of codecs for compressing data, protocols for scheduling transmissions and
feedbacks, and reconstruction schemes for recovering lost data.
If you need copies of papers that are not available in PostScript or PDF,
please send e-mail to
.
Other related work on machine learning
at the University of Illinois, Urbana