Mei Kobayashi has been a researcher at IBM's Tokyo Research Laboratory in Japan since 1988, working on mathematical problems within several different groups at the company. She's investigated inverse problems for the Foundations of Systems (Math and Computer Science) Group, air flow simulations for the Advanced Graphics Applications Group, and acoustical signal analysis for the Media Systems Institute, where, collaborating with a speech group, she helped develop a wavelet transform-based analysis method that is used to make better speech synthesis units for a Japanese text-to-speech software system. The wavelet technique has been adopted by IBM Japan for product development and has also been patented by IBM in the U.S. and Europe.
Mei's current assignment in the Exploratory Network Group is to develop data hiding methods, i.e., techniques for encoding supplementary information in host media, such as digital images or acoustical data. (Data hiding should be distinguished from cryptography, which focuses on the development of highly secure code-decode algorithms and does not necessarily require the embedding of information.) Applications of data hiding include intellectual property rights management (proof of ownership, assurance of content integrity), the detection of geometrical transformations (shifts, rotations, flips, re-scalings), labeling (feature location, image and audio captioning), and cosmetic labeling (visible watermarking using a company logo or symbol). Data hiding techniques are very much needed these days, Mei explains. "The advent and proliferation of digital libraries and World Wide Web-based applications on the Internet have caught legal experts, systems managers, and security specialists by surprise."
"The best part of working in industry is the abundance of problems - like data hiding - which desperately need solutions," says Mei. "These problems are of a completely different genre from those found in textbooks and homework assignments". For example, a product development team doesn't really want the best answer; the team wants an answer which can be found in real time, gives a reasonably good approximation, and is inexpensive to implement or incorporate in a product. Industrial problems are also challenging because they must be solved while satisfying real-world constraints, such as robustness with respect to a number of factors, for example, dirt, measurement error, and temperature changes.
Mei's undergraduate major, at Princeton University, was chemistry, but her love for solving problems ("like working on a sophisticated puzzle") led her to pursue a master's degree in mathematics and then a Ph.D. in applied mathematics at the University of California at Berkeley. As the end of her graduate studies drew near, she sought help from the university's career counseling center, which helped her put together a resume and cover and follow-up letters. Because she had enjoyed an industry internship, she applied for jobs at several companies and received prompt responses. "The interview process was a good opportunity to learn about jobs outside of academia," she says, "although companies do indeed try to glamorize their offices for visitors and prospective employees." When she was offered a job at IBM Tokyo, Mei accepted because it "sounded interesting," the salary was attractive, and she had not yet had begun applying to universities for a job.
"I sometimes wonder what it would have been like to do a standard postdoc and look for an academic job," Mei admits. "The flexible schedule and the excitement of being in a classroom with young people are attractive." But her work at IBM Tokyo, which encourages affiliations with universities and attendance at academic conferences, has actually led to some teaching. She has taught courses on wavelets (at the University of Tsukuba) and on matrix computations (at the University of Electro-Communications). In April 1996, she became a visiting professor of mathematical sciences at the University of Tokyo, as part of a new program in which industrial scientists serve as teachers and thesis advisers.
Mei also considered majoring in English at one time and has turned this interest into another facet of her career, writing science articles with a human interest dimension for an audience wider than fellow researchers in her field. Since 1994, she has regularly contributed articles on Japanese mathematics and on living and working in Japan to SIAM News. "Writing for SIAM News has been a lot more work than I ever imagined," she says, "but the kind and thoughtful letters I've received from readers have made the experience worthwhile."
