Microsoft Research Connections Blog
Next at Microsoft
Social Media Collective
Windows on Theory
Posted by Peter Lee, managing director of Microsoft Research Redmond
The last in a series of posts from the directors of Microsoft Research’s labs worldwide, this one from Peter Lee of Microsoft Research Redmond.Given that this was my first full calendar year at the Redmond lab of Microsoft Research, it took quite an effort to comprehend fully the breathtaking scope of ideas and projects carried out by the lab’s more than 300 researchers and engineers. What a hoot! It’s a great pleasure now to reflect a bit on the past year.This year I used the following “map” to organize my thoughts:
For me, this diagram is a constant reminder of the wonderful diversity of computing research. It also serves as a visual aid to help explain more about the nature of research to product developers, business leaders, and policymakers. The x-axis starts from research activities directed at short-term payoffs and flows out to research that demands greater patience. The y-axis depicts the directedness of the research. It starts from research performed in reaction to an external problem and goes up to the more classical search for truth, understanding, and beauty. Dividing this space into quadrants, we get research that is mission-focused (directed toward solving someone’s immediate problem), sustaining (continuously improving what we do), blue sky (being purely curiosity-driven and often foundational), and disruptive (seeking to surprise or otherwise “change the rules of the game”).For me, the quadrants capture the beautiful diversity of computing research.*
So let’s go “around the horn” by visiting each quadrant and reviewing just a few highlights of interesting things that happened at Microsoft Research Redmond in 2011.Starting with the mission-focused quadrant, this was a year in which the lab worked hand-in-hand with every single product division of Microsoft. Some of these collaborative efforts resulted in products and services that today reach millions of people. What I found particularly interesting were those efforts that applied bleeding-edge ideas in machine learning to solve core problems. For example, the wonderfully effective soft keyboard technology in Windows Phone 7 used machine-learning techniques developed in the lab to understand the intent of even the most “fat-fingered” phone typist. In the company’s big push for NUI—Natural User Interaction—the lab’s researchers provided a foundation of machine-learning research that helped make products such as Kinect and the Microsoft Touch Mouse into critically acclaimed market winners. And the year’s deployment of machine learning in the Amalga health-care system promised to make life better and save potentially billions of dollars by significantly reducing the rate of rehospitalization of discharged patients. Importantly for us, the work on these products also represented great collaborations, not only with product groups, but also with other Microsoft Research labs.Nowhere is it more obvious that machine learning has “come of age” than in our collaborations with the Bing team. This was a year that marked stunning advances in search quality, through techniques such as adaptive search, plus dramatic improvements in ranking technologies, including a first-place win in the Yahoo! Learning to Rank Challenge. And the contributions to Bing didn’t end there; this year, the StreetSlide image-processing technology that we published in SIGGRAPH 2010 also went live in Bing Maps, making it possible for anyone to get wonderfully fluid street-level imagery. These efforts and many, many others resulted in a slew of publications in major research symposia, as well as tangible benefits for consumers.In the sustaining quadrant, few research areas are more bread-and-butter for the lab than software-engineering technologies. For me, one of the highlights was a major advance in so-called “fuzz testing,” a key method for tracking down hidden security exploits in software systems. This year saw the successful internal deployment of a system called SAGE, which uses deep program analysis to scale up dramatically the effectiveness of fuzz testing for Windows and hundreds of other systems. It tickles me that SAGE is the world’s single-largest application of automatic theorem-proving technology! Also of note is the emergence of empirical software engineering, which collects and analyzes huge amounts of data gathered from the activities of software developers to gain visibility and predictive power in large software-development projects.While we are on the subject of software development, it is worth highlighting the amazing fun that can be had playing with TouchDevelop, which enables you to write amazing cloud-powered mobile-phone apps on your Windows Phone. (One-thumb hacking, anyone? :-)) Seeing not only professional programmers but also middle-school kids have fun with this is pretty inspiring.The year 2011 was also memorable for the huge progress made by Microsoft Translator, which gained parity with the competition in translation quality and saw the number of daily translation requests jump by more than 600 percent. Today, we help power the translation services for Bing, Office, and a host of other products and services.When one thinks of the blue-sky quadrant, one often thinks first of research in computer-science theory, and, indeed, there were a number of notable advances in the lab’s theory research. Perhaps the most fun was the work on maximum overhang, a classic problem in mathematics involving how to stack blocks on a table to achieve the maximum possible overhang. The lab’s work on this problem, which was done in collaboration with colleagues from several other universities and industry labs, won the David P. Robbins Prize. This was also the year when the company’s efforts in topological quantum computing, called Station Q, came under the management of the Redmond lab. Based on truly beautiful underpinnings in mathematics and theoretical physics, this year’s research saw a significant ramp-up in companion experimental efforts at several partner universities—and a new architecture/engineering effort in Redmond. And while we are talking about beautiful things, the Buxton Collection exhibit of 30 years of interactive devices was a smash hit during this year’s ACM CHI Conference on Human Factors in Computing Systems.Finally, we have the disruptive quadrant. We were extremely proud to have contributed some of the research foundations and technology for the Kinect audio array. This is a technology that borders on magic, as it enables the Kinect device to hear clearly a person’s voice from across a noisy crowded room, without a push-to-talk signal. This work also shows just how important it is to embrace the diversity of the four quadrants, as more than a decade of blue-sky research contributed to the creation of this disruption in the video-game-console business.In this and many other ways, this was the Year of Kinect. Not only did we collaborate closely with the other Microsoft Research labs and the Kinect product team, but we also flexed our engineering muscles to develop and help ship the software-development kit for Kinect for Windows. This was a total team effort, involving not only the development of the code, but also some beautifully crafted web and media resources and a strong partnership with Microsoft Research Connections to help hackers everywhere get started making cool Kinect-enabled applications.That’s just a tiny slice of the goings-on at Microsoft Research Redmond in 2011. There is really so much more to say, and as I look ahead to 2012, I see even greater potential for major impact. There are some particularly exciting developments brewing in a wide span of areas, such as cloud computing, NUI, mobile computing, entertainment, security, and many, many others. Some of these may involve our new partners at Skype, and then there are some really interesting collaborative possibilities with other industrial research labs. From what we can see, the world’s Ph.D. programs are producing another bumper crop of great new computing researchers; we look forward to hosting hundreds of new interns, postdoctoral students, and visiting researchers in the coming year and will be looking closely at all of them as potential future members of our lab.Of course, none of this would be possible without the great people in the lab. Behind every one of the projects I highlighted above is a team of brilliant people working together to make something great happen. Part of the magic of Microsoft Research is that these people get recognized not only inside the company, but oftentimes in the academic research community. I won’t list the names here (for fear that I will forget someone :-)), but by my count, six of the lab’s researchers this year were awarded “test of time” awards by major research societies or conferences. One was elected to the National Academy of Engineering, and several others were named either ACM or IEEE society Fellows. The list of “best paper” awards is simply too numerous for me to keep track. All of this contributed to a massive amount of interest by people to join the lab. In the end, after sifting through hundreds of applications, we were able to hire more than 20 new people in 2011, some of whom, I am sure, are destined to become the most important technologists of the future.Every day this past year, I walked through the hallways in Microsoft’s Building 99 and rubbed elbows with some of the greatest minds to have walked the planet. Given that, I am pretty certain that, as eventful and successful as 2011 was, the best is yet to come. I can hardly wait.
*This is similar to the diagram by Donald E. Stokes, depicting the concept of “Pasteur’s Quadrant.” In contrast, my diagram attempts to capture the sense that computing does not distinguish between “basic” and “applied” research in the same ways that the physical sciences do.
Very impressive, go Microsoft !!!!!!!
I can see the sky as the limit
Why does Skype run on every competitor smartphone except Windows Phone? Why does every competitor smartphone able to sync with Outlook except Windows Phone? Disappointed with Windows Phone.
thank all of you at microsoft and happy new year
GREAT RESULTS. IT SEEMS LIKE YESTERDAY WHEN I WAS JOGGING NEAR YOUR HEADQUARTERS IN REDMOND IN 1993.
nice...thaks all of at microsoft and happy new year..