BEIJING
, Nov. 4, 2003 — Call it a case of reverse psychology, unintentional or not.
MSR Asia researchers with (front, L to R) Harry Shum, Ya Qin Zhang, Hong-Jiang Zhang. Click image for high-res version.
When Microsoft opened its research lab in Beijing in 1998, researchers there were given the same instructions as those at the companys other labs: Dont watch the clock, and only occasionally glance at the calendar. Put simply, the goal of Microsoft Research (MSR) is to give its researchers the freedom to explore technologies that interest them and that will advance computer technology in the next five to 10 years.
But today, as Microsoft Research Asia celebrates its fifth anniversary, technologies that were advanced and developed at the lab are already well entrenched in a broad cross-section of Microsoft products. More than 70 different technologies developed by MSR Asia researchers have made their way into everything from Xbox games to the Tablet PC to the new Office System. On average, MSR Asia is transferring more than one new technology into products each month.
Just as profound as MSR Asias rate of technology transfer, Microsoft officials and observers outside the company say, is the impact of the lab on the burgeoning research community in Asia and throughout the world. Papers authored by MSR Asia researchers are flooding into advanced technology conferences and journals, with more than 750 published so far. That equates to more than four published papers for each of the 150 researchers based at MSR Asia.
The Asian lab also collaborates on joint projects with universities and other research organizations in China and across Asia, and provides training and mentoring opportunities to hundreds of students each year. Moreover, Microsoft funds research at numerous universities around Asia, dedicating 20 percent of the labs annual budget to projects intended to advance computer science broadly, not just in areas in which Microsoft is specifically interested.
What has been accomplished shows that, with the correct people and correct environment, very impressive results can be obtained in a very short time, says Ramesh Jain, a respected researcher at Georgia Institute of Technology in Atlanta. He also serves as chairman of the Association for Computing Machinery Special Interest Group on Multimedia (ACM SIGMM), a group within the international ACM organization that promotes technology advances within one of MSR Asias target areas of research. The group has selected numerous MSR Asia research papers for publication at the annual ACM SIGMM conference.
The extent of the labs accomplishments surprises even those who are most closely associated with its success. In no way did I anticipate the lab would bloom in the way it has as fast as it has. The contributions and talents have exceeded my expectations in every way, says Ya-Qin Zhang, the Microsoft labs managing director, who was among its first hires. I knew this would become a world-class lab, but not in such a short time.
Technology Transfer Blossoms in Unfettered Research Environment
The rapid success of the lab can be directly attributed to Microsofts support and the talent that the lab has been able to attract, Zhang says. Microsoft continues to increase its annual spending on research and development overall, and plans to dedicate about US$6.8 billion — which is equivalent to 19 percent of projected annual revenue — in 2004.
The primary goal of MSR isnt to invent new products for Microsoft or to pigeonhole researchers into specific areas of research, but rather to move computer-related research and technology forward. Even if a technology developed by MSR doesnt make its way into a current Microsoft product, it may advance the field of computer technology and lay the ground work for future products.
This combination of financial support and research freedom has helped the lab assemble what media accounts have described as a dream team of talent, including some of the worlds top researchers and rising stars in the areas of computer graphics, communication, media compression and content-search technologies.
The research freedom also has allowed MSR Asia to expand the scope and impact of its research far beyond the labs founding mission of developing user-interface and other technologies for Asian computer users. Although most computer users, regardless of where they live, may never recognize many of the 72 MSR Asia technologies that have made it into Microsoft products, some of these breakthroughs are expanding the possibilities of the PC.
“Made” in China: AutoMovie
The AutoMovie features within the Microsoft Windows XP operating systems MovieMaker 2 software are a prime example.
Hong-Jiang Zhang, senior MSR Asia researcher and assistant managing director of the lab, developed software with his team that solves a problem faced by millions of video camera owners who amass endless hours of footage during the holidays, birthday parties or vacations: how to edit the out-of-focus, jiggly and other extraneous shots.
The Microsoft product team for Windows MovieMaker had for years wanted to offer amateur video makers an easier way to create home movies. Using technology from MSR Asia, they were already able to split a long video into shots that the user could assemble easily on a storyboard. But the team wanted to go much further. They envisioned an automated way to select and trim shots, and a way to set it all to music.
It was a tough problem. Zhang and his team came up with a solution: by carefully studying and classifying amateur video and audio, they were able to differentiate good footage from bad. For example, they learned that when an amateur pans across a scene and then pauses while holding the camera steady, that generally means that the shooter probably found the subject matter on which he or she paused to be “interesting.” The team also developed algorithms that can analyze music and find the beat. By combining these two technologies, the team was able to edit the “best” shots from a source video tape down to exactly match the length of a song, and with transitions between clips that matched the tempo. The end results surprised everyone working on the project.
Ian Mercer, product unit manager for Windows MovieMaker, was sold on the new technology after he tested it on a really bad video his friend shot one holiday season. AutoMovie compressed 40 minutes of rambling footage — children running around, dancing, and sitting on Santa’s knee — into five minutes of great-looking video that uncannily spotlighted the friends child, rather than the other children in the production.
AutoMovie has become one of the most popular features of MovieMaker 2 among consumers and technology reviewers. Before MovieMaker 2 we were losing product reviews to the competition. Now we are winning them, Mercer says, crediting AutoMovie with a key contribution to that success. “AutoMovie” isn’t just unique and fun to use, but it also helps reviewers understand that MovieMaker is all about ease of use. It’s not about taking high-end video-editing tools and dumbing them down for the average consumer, its about radical innovation to make the whole process easier, Mercer says.
MSR Asia Helps Make Virtual Ink a Reality
On-screen handwriting whats been dubbed inking — has had a similarly profound effect on mobile computing since its debut last year in Tablet PCs, a new form of mobile computer for which Microsoft developed the underlying software. It is now possible to create, capture and reuse handwritten notes, graphs and drawings on the screen of a mobile PC, using a pen-like stylus.
Much like auto editing was for the MovieMaker product team, inking was a perpetual challenge for developers of the Tablet software. A research team, including several former MSR researchers, set out to provide the user complete freedom to write or draw anything on any part of the screen, not just legible handwriting of a specific size on specific parts of the screen.
After more than 10 years of development, the Tablet team, based in Redmond, Wash., achieved its goal in part because of the contributions of MSR Asia Research Manager Jian Wangs Multi-Modal User Interface team. Among the challenges tackled by MSR Asia researchers was how to better analyze and classify virtual ink. For the PC to properly process and store the virtual ink, the operating system needs to know if the user is writing a note to a friend or drawing a smiley face.
The MSR Asia researchers created algorithms that analyze every stylus stroke on the screen, computing the millions of ways they might be combined to determine if information on the screen is writing or a drawing. If the information is writing, the strokes are sent to the handwriting-recognition engine. Drawings need no additional processing.
Wangs team also worked side by side with the product team to increase understanding about the overall structure of virtual ink on the page, helping refine algorithms that allow the operating system to track writing on the screen even if it slopes down the page or is written in a circular pattern.
What differentiates the Tablet PC from other on-screen writing technologies is the freedom it provides users to communicate however they choose, says Terri Chudzik, lead program manager for Tablet PC. If the Tablet couldnt differentiate writing from drawing, the users options would be limited. So the contribution of MSR Asia was vital to the success of this the new technology.
Getting Innovations from the Lab to the Consumer
Developing new technologies and getting them into products are two vastly different challenges. One takes technical know-how and skill. The other takes business and marketing sense, perseverance and — at least within Microsoft — an affinity for electronic communications.
The MSR Asia researchers and members of the Redmond-based product teams chuckle when they try to calculate the number of hours they spend on e-mail and video conferencing, trading ideas and solving problems.
MovieMakers Mercer says hes been most impressed by how MSR Asias technologies are often close to being ready for inclusion in products. They always think about how to make the technology most useful to consumers, he says. Sometimes that means compromising the overall style or power of the technology to provide ease of use or speed. They know how to create a practical balance, he says.
Mercer and TabletPCs Chudzik were similarly taken aback each time their teams would gather with the MSR Asia researchers. Without fail, the researchers would have something new and unexpected to share a prototype for a new technology, a possible new feature, a previously unrecognized problem they needed to collectively solve.
This input was particularly important with developing technologies, such as inking. They are always offering new ways to think about or use virtual ink, Chudzik says. It helps you break away from your preconceptions and be a little more creative about what is possible.
You need to try to think ahead of the product teams what they need now, what theyll need in their next release, says Hong Jiang Zhang, known for his determination. You have to say this technology is my baby, and I want my baby born.
Another factor in MSR Asias success transferring technology is structural. The lab maintains a team of developers whose job is to meet with product teams to assess their needs and then work with the researchers to find and help develop prototypes for fledgling technologies that meet these needs. This team also provides extra hands and brainpower when other researchers need help meeting a product teams deadlines. Other MSR labs have since created similar teams.
Taking this approach a step further, MSR Asia today unveiled its new Advanced Technology Center. The center will focus on further developing innovations produced in the lab, enabling MSR Asia to capitalize on a larger number of technologies and accelerate technology transfer to Microsoft product groups. In addition, the center will work to create products for Asian users that will improve human-computer interaction. Over the next year, the center will grow to roughly 80 developers and engineers, with plans for further growth in the years ahead.
The Next Five Years
MSR Asia researchers arent worried about their ability to build on the success of the first five years. Neither is MSR.
We built a lab in China because we wanted to have research develop and expand from within Asia, feeding off of and helping grow the vast pool ofcomputer technology talent around the entire continent, says Rick Rashid, Microsoft senior vice president for MSR. This regionalized approach allows MSR Asia to develop technologies that best meet the unique computing challenges of Asia, as well as stretch the boundaries of technology in ways that havent yet been conceived.
Another good omen for MSR Asia and the entire Asian technology industry is the vast amount of young talent in the region. When Ya-Qin Zhang visits Chinese universities, more than 1,000 students will attend his talks. The lab receives 50 employment applications a day and so many student internship applications we just cant count them, Zhang says. Microsoft is helping develop this talent through research grants to Asian universities, internships and fellowships at the lab. Some of these students eventually get jobs at the lab. Most of the rest easily find jobs with other technology firms and ventures, thanks in large part to the Microsoft experience on their resume, Zhang says.
Zhang says the atmosphere at MSR Asia is unlike any research facility hes ever been in, including well-established labs in the United States where he previously worked.
You just feel the energy flowing in the air when you talk with the researchers, with the students. People are really enthused, energized and passionate, Zhang says. It feels like there are no limits on what we can accomplish.
