Dr. Sharad Seth, a professor in the CSE department since 1970, will retire at the end of the semester. His career achievements have been impressive and his teaching will be greatly missed. He speaks about the history of the department, his research interests, and what impact teaching has had on his personal life.
Bits & Bytes (BB): Dr. Seth, congratulations on your upcoming retirement. You have been with the CSE department since 1970. How has the department changed over the years?
Sharad Seth (SS): It has changed in radical ways, both in its size and mission. The department was established in 1968 with just three faculty members on board. Two years later when I joined, we were six and I was among the only two who had any formal training in computer science and engineering. Throughout the 1970's, the department was seriously understaffed, since every course had to be taught to accommodate advanced students. Thus, teaching loads were high and there was little time for research. Contrast this with the 25 full-time faculty that we have now, almost all trained in core areas of computer science and engineering, with lighter teaching loads, and all engaged in research and professional activities contributing to the advancement of their fields. One thing that has not changed, however, is the friendly atmosphere among the faculty and the staff.
BB: Your research and publications are extensive. What has been your most rewarding project(s)?
SS: In my experience the rewards have come primarily from long-term collaborations in new areas of research leading to the projects. I will highlight just two collaborations here to keep it short.
The first started with an invitation from Drs. Vishwani and Prathima Agrawal to spend the summer as a visiting member of the technical staff at the Bell Laboratories in Murray Hill, New Jersey. During my summer visit, we developed a statistical model to analyze the testing requirements for large-scale integrated circuits. In manufacturing, the quality of a test is determined by the fraction of bad devices escaping the test. However, the test-escape rate is nearly impossible to measure with any degree of confidence, so test engineers use a proxy, the coverage of modeled faults, which can be computed exactly through simulation tools. Our model tried to answer the key question of interest in semiconductor manufacturing: what fault coverage is sufficient to achieve a certain test-escape rate? Surprisingly, the n0 model that we developed in early 1980's is still in use at companies like Intel and Qualcomm. My summer work led to over 20 years of a very active and productive collaboration with Vishwani Agrawal, mostly carried out over the phone.
My second long-term collaboration is with Professor George Nagy who was at UNL from 1972 to 1985, and served as the department chair during the first eight years. He introduced me to an entirely new area of research, document image analysis. DIA was just then getting established as an engineering discipline due to rapid advances in the underlying technologies of scanning, storage, computing, communications, and information retrieval. The early focus of DIA research was on optical character recognition (OCR) but Nagy and I wanted to look at the bigger problem of recognizing not just the characters and words but also the document structure in terms of logical components like the document title, headers, footers, paragraphs, tables, charts, images, and so on. The logical structure of a document is closely related to its physical layout; hence we looked at extracting the physical layout of technical documents and proposed the XY-tree structure to represent it. This representation turned out to be very efficient for processing large document images, because instead of analyzing the image in two dimensions, the XY-tree allowed processing alternately in the linear X and Y dimensions. For this reason, it has been adopted in many OCR systems and our 1984 paper is among the most cited of my publications. Over the past 30 years, I have worked with Nagy and others on analyzing many other kinds of documents, including engineering drawings street-maps Chinese and Arabic scripts, and web tables.
BB: Can you tell us any projects or research that you are currently working on?
SS: Professor Hong Jiang and I have wanted to work together for a long time, and it finally happened when we started co-supervising doctoral student Dongyuan Zhan in 2007. Like my previous responses about the collaborations mentioned earlier, this too has opened up a new area of research for me. We are working on managing the cache memory in chip multicore processors (CMPs) for efficient use. CMPs are relatively new and computer architects are exploring different ways of sharing the last-level cache among the multiple cores on the chip. More recently, with Professor Witty joining our effort, we have extended the scope of the work to include the interaction between computer architecture and higher levels of system abstraction, such as the run-time environment, to improve the overall performance of a computing system.
BB: How did you initially get into computer science and engineering?
SS: During my undergraduate study in India, I would read about all the exciting developments taking place in computing, hence choosing the field of study was not a difficult decision. Luckily, after graduating with an electronics and telecommunications degree in 1964, I gained admission to the Master's program at the Indian Institute of Technology, Kanpur, which was an exciting place to be for graduate work for several reasons. Under the Kanpur Indo-American Program many young Indian professors, trained in the US, were returning to join as faculty, bringing with them idealism and the enthusiasm for building this new institution. A large number of well-known US professors also came to the campus for an extended stay. For example, among my mentors at Kanpur was Arthur W. Burks from Michigan, a co-inventor of ENIAC, the first electronic computer. It also helped to know that I was going to join a place that had the most advanced computer in India at that time, an IBM 1620 that you signed up to operate and program all by yourself, in an air-conditioned room – a big deal in India at that time.
BB: What brought you to UNL? If you had to summarize your overall experience at UNL and in the department, what would you say?
SS: A letter of invitation from Professor Lester Lipsky, the Department chair in 1970, passed along by a friend’s friend to me at the University of Illinois, as I was finishing up my PhD there brought me to UNL. I did not know much about Nebraska at that time but was quite impressed by what I saw during my campus visit. The department, working closely with the computing center, had just acquired a large number of state-of-the-art CRT terminals through an NSF grant, placing it well ahead of most US institutions in this regard. The faculty, although small in number, were all passionately interested in the new field and provided a very supportive environment for a new member of the faculty like me. So did the very dedicated staff that the department has had the good fortune of attracting and retaining.
With our growth in size and programming, the faculty and staff may not be as cohesive now, but we still get along very well and continue to support each other.
BB: What, for you, is the most rewarding experience of teaching?
SS: It is most gratifying to receive letters of appreciation from former students, years after their graduation, recalling that what they learned in classes that I taught has served them well in their careers, or that my guidance of their research remains significant to them. I also find working with students one-on-one either on class material or research very rewarding.
BB: What has been your experience on the JETTA Editorial Board? What impact, if any, did this have on your academic career?
SS: The board includes many professional and personal friends. Formally, we meet only once a year at the premier conference in testing. Otherwise, we work independently on reviewing papers. Therefore, the direct impact of board membership on my career has been quite minimal. However, together the board members’ contributions have largely defined the state-of-the-art in hardware testing, which has, naturally, had an impact on my own research.
BB: You were the Director of the Center for Communication and Information Science. Can you tell us a little bit about that?
SS: The Center for Communication and Information Science (CCIS) was one of the first to be funded under the Nebraska Research Initiative, a unique investment by the state to provide a research base within UNL to promote economic growth. To achieve this goal, CCIS provided seed grants and matching funds for federal grants, sometimes in collaboration with private and public agencies in Nebraska.
Unlike other NRI centers, CCIS had a broader focus, encouraging information-technology research by faculty in any discipline. I served as the center director for seven years, from 1996 to 2003. During that time, CCIS associates collaborated with ten Nebraska businesses and government agencies; received many honors, awards, and patents; organized workshops and conferences in document image analysis, cryptography, combinatorics, and coding theory; and established the College of Arts & Sciences’ Area of Excellence in Discrete and Experimental Mathematics. They also helped multiply annual internal funds more than three times with over thirty external grants. In addition CCIS organized colloquia on current topics such as electronic commerce, digital libraries, and quantum computing, bringing many prominent speakers to the campus.
BB: What lessons have you learned from your professional life to apply to your personal life?
SS: 1) I have learned by collaborating with my colleagues and students that every one is different in temperament and work style. When I have not been sensitive to these differences, it has been reflected in the outcome.
2) I have come to realize that although the student-teacher relationship appears to be inherently asymmetrical, I benefit greatly from interacting with my students and am inspired by them as much, perhaps, as they are by me.
3) I have found that new ideas are more likely to arise when working in person than over long distances (although the new collaborative technologies may eventually erase this difference). Furthermore, the most creative results seem to occur when only two individuals are involved and are attuned to each other’s thought processes as well as their pitfalls.
BB: Do you have any advice for CSE students?
SS: The headlines of the day may seem grim but keep in mind that it is a privilege to be studying a field that offers limitless possibilities for improving not only personal fortune but also that of society at large. However, this is only possible if one never ceases to learn. For me that has indeed been another privilege in my life as a computer engineer.