Benjamin W. Wah
September 1, 2008
Q: What is your area of expertise?
A: My current areas of interest are nonlinear optimization, multimedia signal processing, and artificial intelligence. In the past, I have worked on parallel processing, discrete optimization, and data engineering.
Q: Give me a brief synopsis of your education and career.
A: I received my undergraduate and master’s degrees in electrical engineering and computer science from Columbia University in 1974 and 1975, respectively. In 1976 I went to the University of California, Berkley, and received a master’s in computer science. In 1979 I completed my Ph.D. After graduating I taught at Purdue for six years and then came to Illinois. I’ve been here since then.
Q: You’ve had a long affiliation with Illinois. Why?
A: The environment here is very conducive to high-quality research. There are many opportunities to cooperate with our excellent faculty members who are experts in various areas and to find new and challenging areas of research.
Q: Why did you become an engineer?
A: Since I was young, this is what I wanted to do. I have always loved problem solving and mathematics. Electrical engineering and computer science is ubiquitous and involves technology that drives much of today’s industry. It exists everywhere. The profession is exciting because the technology is applicable in so many different fields.
Q: Tell me about a research accomplishment you’re proud of?
R: My group has developed a new approach that breaks massive, complex nonlinear constrained problems into small, manageable sub problems. Using our theory, we were able to glue the solutions of sub problems together in order to easily find a consistent solution across all the constraints.
One of the applications of this approach is in the design of solvers for large-scale nonlinear optimization problems in mathematical programming. With our new solver, engineers can now tackle large-scale problems with as many as 200,000 variables and constraints, compared to the 5,000 that was standard before. This system will have a tremendous impact on solving large-scale problems because the amount of time needed to solve a problem is reduced exponentially. Instead of solving a large problem in a week, it now can be completed in seconds on a PC.
Another application of our approach is in the design of planners and schedulers in artificial intelligence. Our planner won my research group first place twice at the International Planning Competitions, which were held in conjunction with the International Conference on Automated Planning and Scheduling. Our group tackled a collection of major planning problems every ten days for eight weeks using our planner. In the last competition, we successfully solved 91 percent of the problems, while our closest competitor only solved 40 percent. This work is important to many engineering fields, including planning large-scale problems in deep-space exploration and satellite planning for NASA. The system is also applicable to financial markets, manufacturing, and industrial engineering.
Q: What do you enjoy most about teaching?
A: It is an opportunity to work with students and enhance myself. Seeing students flourish is a gratification by itself. I also learn more through teaching the material and interacting with students than I would from simply reading a textbook.
Q: What role do students play in your research?
A: My students are partners in my research. We address problems together and then work on ways to solve them. My interaction with students is an exploration of new ideas rather than a lecture.
Q: What are you focused on today?
We are focusing on problems that are considered challenging in industry, especially those with a high complexity and large size. Solving such problems is not as straightforward as tweaking existing methods. They require a deep understanding of the limitations of current technologies and the application requirements. For example, in a financial engineering problem we are solving, a lot of data is contaminated by noise and as a result, it is difficult to solve the problem in a straightforward fashion. A lot of pre-processing and understanding of the data is needed in order to solve this problem using the methods we have developed. Another application is a voice-over-IP system. The system cannot be designed analytically because many of its techniques require subjective human evaluations and time-consuming experiments.
Q: What does the future hold?
R: The future looks very promising and there are many exciting and challenging application problems to be solved. Since many of these problems have strong impacts in society and industry but cannot be effectively solved today, finding good solutions will unlock the potential of hardware technologies available today.
Q: What technology that’s currently under development are you most anxious or excited to see completed?
In the next few years I will be anxious to see the development of a solver designed for large-scale non-linear optimization problems. The solver will integrate existing solvers to solve smaller problems. We are also working on a new voice-over-IP system that will lead to better subjective conversational quality than existing systems.
Editor's note: media inquiries should be directed to Brad Petersen, Director of Communications, at email@example.com or (217) 244-6376.