OUR STARS 2000

Jerry Xiaojin Zhu
"In five years, speech recognition technology will be more integrated into people's everyday lives."

Roni Rosenfeld (foreground), Jerry's advisor,
with Jerry Zhu (background).

LTI PhD Student Wins Microsoft Graduate Fellowship at CMU

Microsoft Research has awarded its prestigious Graduate Fellowship to Carnegie Mellon's Jerry Xiaojin Zhu, a 2nd year PhD student in the Language Technologies Institute (LTI). Carnegie Mellon University (CMU) is one of only eight universities in the country invited to nominate candidates for the Microsoft Graduate Fellowship. Microsoft then selects one recipient from each school to receive the award. The award is generous, paying for tuition and fees, and providing a stipend of up to $20,000 to cover living expenses for one academic year, with the option to renew. It also affords a laptop and Microsoft software of the person's choosing.

Roni Rosenfeld, one of Jerry's advisors, nominated Jerry for this award because "[he] is outstanding in his intellectual ability, thoroughness, scope of interest, and productivity. Even more importantly, he has an unusually mature view of the scientific process." Jerry has been working with Dr. Rosenfeld on statistical language modeling and speech interface design, as well as with Alex Waibel on multimodal integration. Language modeling is part of speech recognition, which involves converting the human voice into phonemes and then reconstructing these phonemes into possible sentences. At first, Dr. Rosenfeld was concerned that Jerry might overextend himself in trying to do so much; however, he soon found that Jerry was contributing to each project as much as a dedicated full time student. Dr. Rosenfeld summarizes Jerry's achievements, saying: "In the little over a year that Jerry has been here, he managed to make progress, in parallel, in statistical language modeling, speech interface design, multi-modal integration and machine learning. All of these resulted in (or are about to result in) publications."

Jerry's research focuses on building statistical models to determine the probability of natural language sentences. He is also pursuing research in multimodalities, an area which looks at the ways computers can receive human input, expanding from keyboards and mouse to speech, gestures, handwriting, lip reading, gaze, etc.

Before coming to CMU, Jerry finished his BS and Masters in Computer Science at Shanghai Jiao Tong University in China. He then went on to work for two years at IBM China Research lab in Beijing. It was at IBM that Jerry first began looking into new ways of interaction between humans and computers. His desire to learn more led him to apply to doctoral programs. LTI offered Jerry the program that best matched the line of research he began at IBM in Beijing.

When asked where would speech recognition be in five years, Jerry surmises that speech recognition will be more integrated into people's daily lives. Home appliances will be designed with speech recognition technology, like what Jerry's working on now, allowing people to vocalize or gesture their commands. Lights will come or turn off and food appliances will make you the beverage or food of your choice, a la Star Trek. Whatever the future of speech recognition may be, it's sure to shaped in some way by Jerry Zhu.

For more on the Microsoft Graduate Fellowship, see http://research.microsoft.com/msrinfo/opps/fellows.htm.

Also see http://www.scs.cmu.edu/~zhuxj/