Microsoft Research and Innovation Fair
Brussels, Belgium
September 23, 2004
ANDREW HERBERT: So I’ll take just a few more moments to talk about the format this morning and then we’ll get into the presentations.
First I’m going to invite Jean-Philippe Courtois, who is the president of Microsoft in EMEA, to talk about Microsoft’s role in EMEA, how as a company we think about and respond to the Lisbon objectives and the broader issues of innovation; then Rick Rashid, who is the Senior VP of Research for Microsoft, will talk about Microsoft Research as a whole, the organization. Rick is my boss. I run the Microsoft lab in Cambridge. I will come on after Rick and talk about the things that we’re doing in our Cambridge laboratory, the relationships we have with many universities across Europe and the collaborative projects.
So with no further ado, I’d like to invite Jean-Philippe to come and tell us about Microsoft in EMEA. Jean-Philippe, thank you. (Applause.)
JEAN-PHILIPPE COURTOIS: Thank you, Andrew.
Good morning. Good morning and first of all I’d like to thank you all for coming here from different places and I’d like to extend a special thanks to the newly elected members of the European Parliament who I know are already and will be very busy by a number of key discussions. I’d like also to thank the Honorable Malcolm Harbor for the great introduction and also many business partners and academia in this room. Together we will join with the EU public sector leaders to truly help build the European future we all know is possible.
I and my company are very excited actually about the opportunity of bringing innovation to the core of Europe. And partnerships are really essential to make that happen. Partnerships happen to be one of the core strategies of a company since the first day we existed as a company.
So today what I would like to talk about in the next few minutes is really the way how through innovation we work to achieve that vision at Microsoft, the work that would enable us to be a great partner to achieve the Lisbon goals, this bold vision that Malcolm was talking about.
And if you look actually at the IT industry and the way this ICT industry has come over for the last 30 years, I think everybody will agree that it’s been an amazing period of time in terms of innovation.
And the reason we are optimistic about innovation in Europe is because we are in the innovation business as a company. We started, as you k now, back in 1975 with this bold vision, people at that time would say a nave vision, to have a computer in every home, on every desk, and, well, right now all across Europe, certainly in Western Europe the PC penetration in homes is above 50 percent, still some way to go in Eastern Europe, that’s true, but obviously Internet penetration at home and the business PC usage is super high. So this is certainly something that helped us realize how much innovation can bring to the marketplace.
When you look also at the adoption rate of the PC industry compared to some other big innovations in the past, thinking about the automobile industry, thinking about the phone industry and electricity, you just realize that when it took five years to get roughly 160 million users on the Internet and 600 million users, PC users, that took probably more than 15 years to do that; it actually took about 80 years to get the same kind of penetration with the automobile industry and about 30 years to do that with electricity. So I think that also helps us to get a sense again of the speed of innovation.
So as a technologist in particular in Europe, I feel and we feel extremely confident that innovation again holds great promise for Europe and for meeting again the Lisbon goals. Innovation not only enhances the way the society works, but it’s also actually a great booster for productivity of the entire economy and, of course, in a direct way fueling the growth of the information communication technology sector.
If you look also at employment, just taking the example of my home country, the ICT sector is accounting for 7.2 percent of the workforce in France, which is frankly something you will see in many other countries, so a very significant impact in the shift of the European economy.
And so today as European leaders, making kind of a checkpoint halfway through the Lisbon agenda, I think actually there are a number of meetings happening next week under the Dutch presidency, reviewing the goals, I think it’s actually a very good time to reflect on where we are, and certainly to reinvigorate and refocus on the process again to get there, to get to those bold goals.
And this is really the kind of commitment we like to take as a company to basically supporting all the work initiative going on in Europe. As a company, and Rick will touch on that, we are investing about US$6 billion in research and development per year worldwide and the way we try to activate the innovation process really you could stretch all that along three different lines. You could think about, of course, all the internal innovation made within Microsoft, and I’m going to touch on that and much more to come later. You can also think and you should also think about the joint innovation we do and we do more and more with some partners. And the third type of innovation is innovation we do as a private sector, the private ICT sector in Europe and in the rest of the world.
So on the research side, again, many details will be provided to you, but we are very excited back in 1987 when the Cambridge facility was set up in Europe to open up Microsoft Research capability in Europe. And today we have I think the top 85 minds, great people — you can actually meet and discuss with some of them downstairs. I hope you have a chance, please do so — who come together to really look at some blue sky projects. They are really looking at long term projects that could have an impact five years from now, 10 years, sometimes probably 20 years from now.
The second type of innovation we want to also accelerate in Europe is actually product development, which is something we do in Denmark where we have a product development center in Vedbaek with 700 people, who are developing business applications for the needs of small and medium sized businesses all across the world. And the fact that this company is located in probably the most dynamic small business economy, which is the European small business economy with 20 million companies, has fueled a lot of great new features in innovation in worldwide products on ERP, supply chain and all kind of e-commerce applications.
The third type again of internal innovation is what we do in Ireland with our European Product Development Center, which is in charge of localizing our products. It’s one thing to have Windows, Office and our products in English, but of course if you want to make that relevant for all of us in the world, they’ve got to talk your language. So this is exactly what our Irish team has been doing for a number of years and they have been actually localizing 100 products in 35 languages and we have recently over the past few years extended that with a new program that we call the Local Language Program to add an additional 40 languages. Last year I was happy as I was in Malta to actually announce the Maltese version of Windows and Office and there are many more actually that have been developed all across Europe and the same also in Middle East and Africa.
The second type of innovation is what I’d call the joint innovation and this is where we started to be much more proactive I think a year ago when we opened up Microsoft Europe Innovation Center in Aachen in Germany, at the core, I would say, of European innovation to basically work on some EU projects, which are the key priorities of the research directions from the EU and working with a number of partners, working with companies like Telecom Italia, working with integrators like Atos, working with SAP, working with universities of Berlin, Politecnico di Milano and many others and working also with governments, actually doing a number of projects right now with the German government as well.
So this is a new type of joint innovation where we plan to bring together with those partners some actually real technologies related to security, related to e-health, related to e-learning, which can really provide some great benefits again to the Lisbon agenda, as defined by the research agenda.
And finally and certainly not the least, I think when you have to think about innovation in the context of a company, we need also to relate to the 300,000 companies in Europe working with Microsoft on a daily basis. I’m talking about all those small, midsized and sometimes big companies such as service providers, system integrators, resellers and I’m also very much talking about the 25,000 independent software vendors, basically software companies like Microsoft, writing code and applications on the Microsoft platform, which have created an incredible healthy sector all across Europe and a number of great applications both in the consumer, in the business space, in the enterprise, small business, et cetera, and we certainly take this responsibility very seriously and you will see us extending in the future even more I would say not only our reach but certainly our partnership with many of those software vendors in Europe, which really have created a lot of very highly valued jobs in many of the European countries and exporting those applications actually all across Europe, some of them, and even across the world.
So altogether those are really the areas of focus we have to be I hope a true and actually a very concrete partner to this innovation charter in Europe.
So, in summary, again I’d just like to leave you with this commitment of our company, to be more and more actually investing in Europe on all the fronts of innovation that I was talking about and we want to do that also as a platform provider. We are committed to provide this technology platform based on open standards, fully interoperable, because this is how our customers are actually getting the systems to connect to each other every day, and we want to do that again in a close partnership with many institutions.
That’s why I think we were very pleased last week again to announce some new developments in our Shared Source Licensing Program as one example. Based on the work we did in Windows, offering access to the source code of Windows, last week we announced that we will do the same in Office and the first actual agreement was signed with the government in the U.K. to have access to the Office source code.
The same thing happened actually last November when in collaboration with the Minister of Science Innovation in Denmark we made an announcement, which is now available for everybody, to have a free-royalty access to the so-called Office XML Schemas to basically make it super-easy for any institution, developer, individual, organization to basically interoperate, exchange documents using XML as a core standard.
And you will see our company doing more and more to again show some very strong commitments about the way we can extend innovation working with both the public sector and the businesses.
Now without further ado, I’d like to introduce Rick Rashid. Andrew did already a little bit of that. So Rick is the Senior Vice President of Microsoft Research. He joined us back in 1981 and although research had already started, I must say because I’ve been in the company, as you may know, now for 20 years, that was still in the early stage, really at the infancy, and I think a lot of credit goes to Rick for really putting research not just on the map at Microsoft, for Microsoft, but externally growing, structuring and I think basically extracting a lot of value from this research capability.
And so a lot of work has been done and I’m sure Rick will not be short on any examples but on operating systems, digital media, data compression and many other things and innovations that actually have happened in our current products for the last 10 years.
Before coming to Microsoft, Rick, of course, had another life and he was a professor at Carnegie Mellon University where he worked on a very well-known Mach operating system, Mach, M-A-C-H. I just want to make sure nobody is confused; this is that Mach, which is still actually in the marketplace today and which had a very big impact in terms of operating system development.
So it’s my very pleasure now to welcome Rick Rashid to really talk about the vision we have for research and probably relate certainly to Europe as well his examples. Thank you. (Applause.)
RICK RASHID: Well, thank you, Jean-Philippe.
I’m really excited about being here and I want to thank you for giving me this opportunity to share with you some of the excitement that we have in terms of the opportunities that technology is going to be creating over the next 10, 15, 20 years in the future and to really show you some of the exciting developments that we have been creating within Microsoft Research and that have been created by some of the universities that are working with us in the field.
So as Jean-Philippe just said, I came to Microsoft in 1991 and really we started then to create a basic research organization within Microsoft. Today we’ve grown to be 700 researchers around the world. And to put that in perspective, we’re the second-largest basic research organization in computer science in industry, and we’re continuing to grow and grow dramatically and really at a time when many other organizations have been shrinking their basic research in industry. So we feel basic research is important to us.
Today we have five research labs. I’ve listed them here. Our first lab was in Redmond, Washington, which is where Microsoft’s corporate headquarters is located, but I think importantly the first place we went outside the United States was to Europe. We created a research lab about eight years ago in Cambridge, England, and that’s been one of our major research facilities and one of the institutions that’s really helped to push forward our research agenda over the last eight years. So that’s been a significant investment for us. We’ve also created labs in Beijing and two other labs in the United States, in Mountain View and San Francisco. So we’ve been growing our research organization over the last 13 years to about 700.
The research lab here in Europe, as I mentioned, really started back in 1996 and we built it in close collaboration with the University of Cambridge and really with the view that we would be a partner with European universities. And so as we’ve built up our research lab here, we’ve also tried to build our relationships with universities. So we’re trying to strengthen our links to the European research community, and really in a way what we’ve done in every environment that we’ve created research facilities is we’ve viewed what we were doing as really creating a technical aspect not just for ourselves but for the organizations around us.
This is our mission statement, this has always been the mission statement, and I think it gives you a sense of who we are as an organization. The number one mission I have within Microsoft Research is to move the state of the art forward in computer science. Now, interestingly enough, I mean a lot of people look at that and they say, but that doesn’t say anything about Microsoft, and it doesn’t. I mean, our goal isn’t really first and foremost to worry about what technology Microsoft will need but rather to say where is the state of the art in the different fields in which we do research and how can we move that forward, how can we change that.
Now, the reason that’s my first mission is my belief is that unless we are moving technology forward, unless we are at the forefront of the fields that we’re doing research, we’re not going to be an asset for a company like Microsoft or for anyone else for that matter. So our first mission is to move the state of the art forward.
Our second mission is, if we do that and we create technology that makes sense, we’ll work really hard to take that technology and move it into our products, and we probably are as good or better than that as any other industrial research lab in the world.
And again I think the key thing is the third bullet point, which is by doing those first two things, my real goal in the organization that I have is to make sure that Microsoft is still around in 10 years. A lot of people don’t really remember back to the old days when Microsoft was really a small company. When I came to Microsoft in 1991 to create the Microsoft research organization, the biggest job I had in terms of hiring people into my group was to convince them that Microsoft was still going to survive five and 10 years into the future.
And really today we seem like we’re a much larger company, it seems as though our existence is assured, but in our business things change very, very rapidly and unless we innovate, unless we continually change, we’re not going to be here in 10 years, we’re not going to be here in 20 years.
So that’s what we’re about, that’s what we’re trying to do, move the state of the art forward and make sure that Microsoft continues to be an innovator well into the future.
Here in Europe we’ve grown an organization now that has about 85 researchers and innovators in Cambridge and we cover more than 17 different nationalities; its’ a very pan-European organization. There are some Americans there as well.
And the goal really that we have is to do research that’s going to change the world, that’s going to make a difference. We bring in lots of students and another part of the way we think about ourselves is we’re also an educational institution. We bring in a lot of interns, we have 50 just in our research lab in Europe, but if you look worldwide we have something like 450 PhD interns that come into our research labs worldwide. Just in Redmond, in our facility in Redmond, Washington this summer we had 184 PhD interns.
And to put that in perspective, in the United States only about 800 or 850 PhDs are produced in a single year, so a large fraction of the PhDs that are produced around the world have probably spent time at Microsoft Research, and so that’s a really important part of the way we think we do our work.
Here are some of the people that we have at our research lab. We have probably one of the most decorated groups of computer scientists in the world. We have three Turing Award winners, one of which is in our research lab in Cambridge, Sir Tony Hoare. We have a number of members of the National Academies in the United States and the Royal Society in Europe. We have individuals that have played significant roles in helping to create the scientific infrastructure that today is computer science.
But one of the things I’m really excited about is we also have a lot of young people and you’ll see that in the conversations you have with our researchers downstairs later. We have a lot of young researchers that are making their careers at Microsoft and really establishing themselves as leaders in their fields. I’ve been really pleased to see how much that’s happened. Just recently we’ve had three of our young researchers listed in the technology reviews list of top 100 innovators in the world. One of our researchers, Hughes Hoppe, was just given a distinguished service for the research that he’s done in computer graphics at SIGGRAPH, so we’ve had a lot of really great young people. We’ve listed some of them here that we have in our research lab in Europe, but it’s exciting to me because that’s really the future, that’s really what’s going to continue the role of innovation.
Now, Jean-Philippe mentioned that before I came to Microsoft I was a professor for 12 years at Carnegie Mellon University in Pittsburgh, and the way I think about doing basic research is the university model and we’ve built Microsoft Research on that model. We’re much more like a university Computer Science department than we really are like a traditional research lab that you might think of is in industry. We’re very open with the work that we do. We publish a tremendous amount and I’ll talk a little bit more about that later. But one of the goals that we have is to share our research with people on the outside and we judge our researchers on their peer-reviewed research, just like a university would.
We also reach out and we try to be part of the academic research community, so we work with universities and more than 15 percent of the money we put into basic research in my organization goes to universities directly, either in the form of external research grants, fellowships for students, support for teaching and support for curricula. And one of the things we’ve tried to do is be a catalyst to help cause innovation in the teaching process within the university systems.
So that’s an important part of what we think of as our mission and the way we interact, and at any given point in time there’s probably something like three dozen visiting professors, people from universities working in our research labs, sharing ideas, sharing what they’re doing with our researchers and our researchers sharing what they’re doing with them.
Now, I mentioned this notion of working and collaborating with academia. One way to look at that is just to look at the kinds of papers we publish. Many of them are joint papers, many of them are papers done with people in universities. Again, we view that as an important part of how we think about ourselves.
We also sponsor and create a number of symposia and conferences, especially in new areas. For example, just this year we sponsored a conference in social computing, really one of the first conferences in social computing, bringing together not just academics but also practitioners and people from industrial research labs and creating a dialogue between them about what we can do with computers that enhance social structures, that allow people to interact together and to share ideas and share their feelings with each other.
We sponsored a conference in mesh networking, which is really a new area of sharing and communicating in wireless networks in new ways, and you’ll see some of that research downstairs later on.
We’ve really tried to take a leadership position in creating sort of new academic areas. We were one of the first institutions to support research in data mining and we had a symposium now more than 10 years ago that really helped to create what is the modern data mining field. So that’s an important part of the way we think about what we do.
We’re working, as I said, to expand the state of the art and one way to see that is in the papers that we publish. Out of our research lab here in Europe we’ve published more than 70 papers in top conferences and journals over the last year. Worldwide we’re probably one of the top publishers of research now in the literature. Just this last summer — or summer isn’t quite over yet I guess although it’s looking pretty dreary out there right now, but just over this last summer, if you go to many top conferences, you’ll see more papers from Microsoft Research than any other institution in the world. At SIGGRAPH we had 13 papers out of 81. We had more papers at SIGMOD, which is a major database conference, than any other organization, more papers at SIGIR, which is the major conference in information retrieval, more papers that will be coming up in something called OSGI, which is one of the major operating system and networking conferences.
So we’re really a major contributor to the advancement of science and we view that as an important part of what we’re doing and the way we measure ourselves, and we view our community leadership as an important part of that and our researchers here in Europe are taking leadership positions themselves in many conferences, many organizations within Europe.
The impact we’re also having on our product groups is extremely high. I started the group which today is the digital media division at Microsoft. I started it as a research project back in 1993. We built it into a product organization in 1996 and we spun that out and now they’re a major part of what Microsoft is doing.
I started the first e-commerce group at Microsoft. Our research lab in Cambridge was really the place where the Tablet PC was born. I can remember visiting that lab in the late 1990s and seeing the researchers there with their laptops cut in half, flipped over, digitizing grids and experimenting with the software that became the basis of what was the Tablet PC.
A lot of what you think of as Microsoft today has come out of the research organization. Something like 10 of our current vice presidents in our product groups are people that have worked for me in the past.
So Microsoft today is really a product in many ways of the basic research work that we’ve done over the years and many of the key innovations that allow us to be successful in the marketplace have come from our research organization.
Here are just some examples that I’m going to be talking about today: some of the work we’re doing, for example, right now to really change the way people think about doing machine translation from one language to another and I’ll show you a little bit about that; and some of the spin-off technologies, things that really aren’t so much going to be Microsoft products but are technologies that other people can use to create products, and I’ll talk a little bit about some work we’re doing in biometric IDs.
So let me talk a second about machine translation as an example of how we’re helping to drive our product activities.
One of the issues that we have as a company, and Jean-Philippe referred to that, is it’s very hard for us to provide our services in all the languages that we need to provide them. One of the big issues we have as a company is translating all the materials that are generated, whether they’re generated in Europe or generated in the United States or generated in Asia and taking them into other languages. We spend something like $200 million a year just translating things from one language to another. And I know there are many versions of documents in this library that are simply translations from one language to another and I think that’s a key problem that a lot of people have.
So what we’ve been trying to do is say how can we try to make machine translation be usable and be able to reduce some of the cost and really improve the experience that our customers have.
One of the things we’ve tried to do is build machine translation services that are data driven, so we can take previously translated documents and use those documents to create a machine translation system that then can be used to translate new documents.
And we’ve been using this to attack a particular problem that we have, which is our knowledge bases. These are large collections of documents, over 130,000 documents that our customers can go to and look at on the Internet when they have a problem or when they want to learn something about what our products do and how they do it.
Now, historically only a very tiny fraction of the documents that we’ve had we’ve been able to translate into other languages. Most of the documents start out in English and then we have to translate them and typically we’ve only been able to afford translating 2,000, 3,000, 4,000 documents out of those 130.
What we’ve begun to do is use our machine translation technology, created within our research group, to translate all of the documents into certain languages. So today if you go to Microsoft’s Knowledge Base in Spanish, we now have all 130,000 documents available for our Spanish speaking customers. Almost all of those are machine translated.
And what’s exciting to us is that we survey, we test to see how well this information solves our customers’ problems and what we’re finding is that the machine translated documents are doing almost as well or in some cases slightly better at solving the problems that our customers have and satisfying them with the information that they’re getting as the human written documents in English are for the English-speaking people. This is in some sense the real test of machine translation; it’s not does it sound right, is it useful, did it do the job, and we’re able to do that with the work we’re doing.
We started with Spanish, we now have Japanese up and running and so our Japanese customers now have access to these documents. We’re running pilot projects right now with our knowledge base in French and German and we expect to roll out more languages as time goes on. And we’re really excited about this because it creates a better experience for our customers and it solves a problem that we’ve had in the past in moving our information.
Here’s just an example of the translated material from one language to another and you get a chance to see the kinds of translations that we are able to do with this system. Again, it’s completely automated and really the goal of this is not to create new language systems every time we wanted to go to a new language, but really to be machine driven, data driven as we go from one language to another, and we’re working with many organizations now to take advantage of it.
Here’s another example of really a spin-off work from the research we’ve been doing in cryptography and some of the work we’ve done in computer vision where we’ve been looking at the problem of how can people identify themselves in a way that’s unambiguous without having to have central databases, without having to have consultation to some computer out in the distance and without having to have a special ID, a special kind of thing that if they lose it it’s a problem and something has to go wrong.
So what we did is we took our research and we created a technology that lets us create a kind of an ID that can have your picture, information about you written on it in a way that cannot be altered because any alteration could be immediately detected by a scanner that would scan this and say, no, that wasn’t what was signed, that wasn’t what was asserted by some authority, whether that’s authority as a government or a corporation.
What’s great about this technology is that you can encode on just simple pieces of paper information about what the person is allowed to do, for example, what access they may have to parts of a building or information about the individual that might be important such as passport types of information.
And yet the document itself is not important. It could be just printed on a regular piece of paper and it would be valid. If it was torn up, you could print another one and it would be just as valid. And only that person could use it, because it has that person’s picture and biometric information on it and it can’t be altered because any alteration could be detected.
So we’re excited about that. It’s not a technology that Microsoft would use directly but we’re in discussion with many partners to license this technology to them and to help them use it to create their own new opportunities in business or in government.
Here are some examples of the advantages.
Another area that we’ve been working in, and you’ll see some examples of this downstairs, is we’ve been trying to say how can we use computer science to accelerate the rate at which other sciences move forward and to really not just innovate in our technologies but really open up innovation in a wide variety of fields and speed the innovation process.
So we’ve been looking at ways that we can work with academia, using our platforms and using our research to accelerate the rate at which other science can be done, whether that’s in things like astronomy, education — SkyServer and Conference XP are astronomy systems and education systems. We’ve been talking to the biological communities. Increasingly biologists are looking at the cell, they’re looking at the body and the genome and they’re realizing that those processes look a lot like what we have been doing research on and the kinds of mathematics we’ve been examining in the field of computer science. And so we’ve been working with them to solve some of their critical problems.
In our research lab in Redmond, for example, we’ve entered into a collaboration with the University of Washington and a number of other institutions in really tackling the problem of finding an AIDS vaccine and really recognizing that some of the key problems there are really computer science problems and that we can use the technologies that we have in machine learning to help winnow out AIDS vaccine candidates and that we can then make them available as a Web service so researchers around the world can have the benefit of being able to do that analysis without having to leave their home countries or even their desk necessarily.
We’ve been looking at things like the early detection of cancer. We’ve been working, our research lab in Europe has been really looking at how can we take some of the technology that’s been developed for parallel computing and the languages and the mathematics of parallel computing and use that to address the question of how does the cell work, how do we solve problems of biology that exist. So we’re really excited about that and the opportunity to make a difference there.
Just a really quick example that I wanted to show you is the SkyServer. Many of you probably know that we put a database up in 1998 of images of the earth’s surface, something we call the TerraServer so people could see their homes, they could see the things that are around them, and it was one of the first large databases of information, in this case from government sources, that could be made available to the broad public. And we received an award from the U.S. government for the work that we did to make the U.S. Geological Service data available to the public.
More recently, we’ve been doing the same thing with the astronomy community, working with astronomers to say how can we take the great data our telescopes around the world are getting and make that data available to scientists all around the world 24 hours a day, create a virtual telescope that never goes dark.
So we’ve been working with the Sloan Digital Sky Survey, with a number of universities and with partners here in Europe to put online this database. This is a picture of the SkyServer and one of the things, if you go to this Web site, one of the things that you’re able to do as a layperson is see some of the beautiful imagery that’s there.
But for the scientist, they can go to this same data and they can find research articles about the phenomena that they’re looking at. They can data mine this information and they can now do new research, they can find new astronomical phenomena much faster than they could before, so this is a great innovation for them.
But even more to the point, it’s a great thing for students and for teachers because now everyone in the world has access to the same data, the same astronomical information that the best scientists in the world have access to. So we’ve democratized the information in an important way. And the entire set of sources and the complete environment that supports the SkyServer is available free to anyone around the world that wants to use it for their own purposes and to do their own work.
I want to mention just as I’m finishing up some of the work that we’ve been doing reaching out to the academic community. As I mentioned earlier, we’ve taken from the very beginning of my organization our work with universities very seriously. We have an external research office in our research lab here in Europe that is really focused on working with universities.
And one of the things that we’re doing right now is we’re taking a hard look at how can we do even more than we’ve done in the past working with universities to really accelerate change and in particular how can we help computer science departments work with biology departments, chemistry departments, physics departments, other disciplines to accelerate science broadly and so we’re putting a lot of effort now through our external research office in doing that throughout Europe and, in fact, I’ve been traveling over the last few days and meeting with some top academics at key institutions in INRIA, a number of other places, really opening a dialogue about how we can even do more as an organization to advance the rate of progress in our field.
And here is some more information about it. These are really some of the initiatives that we have going forward. And I want to emphasize this last one, which is intellectual capital development. I think one of the key things that we can do as an organization is to help create the future by helping create the future scientists that are going to really innovate and create change over the next 10, 15, 20, 25 years.
As I said, I came from being a professor. My whole career has been built around training people and moving people forward and we view that as a critical part of what we’re doing and again you’ll see downstairs during some of the presentations some of the work we’re doing with universities in Europe, whether it’s in Berlin, whether it’s Lyon and some of the universities that are working within INRIA and others. In fact, just this last year we’ve supported research projects in 95 universities in Europe, so we have a very big commitment and we’re making a greater commitment moving forward.
So with that, I am going to now — this is my slide that says I’m really excited about the future but hopefully you’ve gotten the impression already I’m pretty excited about what I do.
I’m going to now pass to Andrew Herbert again and let him talk about specifically some of the exciting things that are going on in our research lab in Europe and some of the things that you’ll see downstairs. Andrew? (Applause.)
ANDREW HERBERT: So Rick has given the introduction; what I want to do in a few minutes is tell you specifically what we’re doing in the Europe lab in Cambridge, introduce you to some of the projects, convey some of the spirit of our research and set the context for the demonstrations, the exhibits that you can see downstairs.
So the first thing I’m going to do is run a video for about five minutes. We made this video just over a year ago and it shows the founder of our laboratory, Roger Needham, who is a very well-known academic, who very much set the tone and the orientation of our lab. You’ll meet a number of the key scientists talking about what it is that excites them as researchers and the things that challenge them in their work.
(Video segment.)
ANDREW HERBERT: So I hope what you got from that video is a sense of the passion that the researchers in Cambridge feel for their chosen areas of research. One of the pleasures of being a lab director is working with people who have that enthusiasm, who have that breadth of knowledge about their subject.
I think the thing which distinguishes an industry researcher from an academic researcher is that drive to see the research go through to the product groups and the product groups into market and see your research have an impact on the world. That’s the tremendous opportunity.
So the things that drive us in the Cambridge laboratory are clearly innovation. We have an internationally recognized research competence in our key areas in Cambridge. We publish internationally. Our scientists are known internationally. Many of them are European or have chosen to come and work in Europe because of that competence.
We look to leverage some of the European strengths in computer science, particularly, for example, in areas like software specification and verification, networking, my own area, machine learning; those are subjects that are very strong in Europe.
Our people, we choose to hire the best who want to work in Europe on our topics of research. Increasingly, people come to the laboratory not just because it’s the lab in Europe in the European context but because of the topics and the challenge that they present.
And, as Rick has said in his presentation, collaboration is extremely important to us in many ways. Our research is open. Most of our projects often have an academic partnership taking part in our projects, whether consulting with us closely on our project or doing work alongside with a professor or through supporting a PhD student, and working together to publish our papers.
As Rick said, we published over 70 papers in the top journals in our field from Cambridge last year. In fact, I think some copies of those papers are downstairs in the exhibition room. You can look at them, there will be a short exam at the end of the day.
Many of those papers, you’ll see the authors are not only from Cambridge, from the Microsoft lab in Cambridge, but from our partner universities and that’s very important to us, developing our own research and stimulating their research as well.
And we’ve talked about the external research office where we’re looking at research at the boundaries of computer science. In the Cambridge lab we look at the core subjects. Through our external research office we’re seeing how computing can reach out to the other sciences, what can computing bring to those sciences to make them more effective, how do their challenges, their need for data mining, their needs for scientific computation, their needs of visualization, what challenges they bring to us as computer scientists and how might some of our scientific methods help them in their science.
Biology is the interesting case. Biologists yet don’t have the mathematical foundations to develop their science in the way that the physicists and the chemists do and we’re finding that some of the mathematics that we’ve developed in computing to describe software, which itself is complex and multilevel, those mathematics are turning out to be useful tools for biologists so they can get hold of biology and describe it, and then that lets them take their science further forward, and we’re very excited by those contributions.
So I’d talk a little bit about the organization of our laboratory. Really it is two parts. There is the laboratory itself in Cambridge, which does mainstream computer science research and then we have our external research office, which builds on our earlier work with universities and is now focusing on this scientific theme.
In our core computer science lab we have four areas of research. If you noticed in the video, we talked about three of them. The fourth one, interactive systems, is something we’ve been building over the last year. We’re very excited about that. We’ve been able to hire some great people to come and join us and open up a new agenda.
In the systems and networking group, which is my area of research, I find it fascinating, I got my PhD in 1978 for writing operating systems and here we are in 2004, there are still challenges in operating systems. They are different challenges, the computers are different and we worry about operating systems in datacenters with tens of thousands of machines, we think about topics of distributed computing, Web Services, those networks; huge, huge challenges still to work on.
And so in my group there are people looking at that from the perspective of the operating system, from the design of the network. You’ll see downstairs one of the demonstrations is looking at how we deal with mobile networks, a concept of mobile hotspots, making those networks much more efficient.
And a very important area that we call distributed computing, this is all about computers working together, interoperability, how can we share data on those computers. Many of us have many machines. We probably have a computer in our office, a computer in our home; we may even have one of these things. You might think it’s a cell phone, I think it’s a computer. How do we manage our data across those things and how do we keep them consistent with each other? What do I want being sent to this as opposed to my Pocket PC? Those are the challenges of building systems.
Our second group, led by Chris Bishop, is our machine learning and perception group. Machine learning is a very important part of computer science. Back when I was doing my PhD in 1978, computing was binary, ones and zeros, yes and no. The big change that I see in computing is it’s now about statistics. In machine learning you’re using statistics to try and carry forward a hypothesis about does one document match another. If you’re doing something in image processing, if you’re trying to extract a feature from an image, the hypothesis might be where is the boundary, is it this line or is it another line. With machine learning, you can represent different hypotheses as different pieces of statistics, manipulate those mathematically and as your algorithms work on the data, they can learn from previous data, they can statistically tell you what they think is the most likely outcome, and that approach to computing is very exciting, it’s the way in which we can tackle those problems that can’t be reached just by programming with the ones and zeros that I worked with as a young research scientist.
So in our machine learning perception group we have a very strong focus on the machine learning itself, the algorithms. Chris Bishop, the leader of that group, is a well-known author in that field. Indeed, he tells me he’s about to publish a leading book on the subject.
Computer vision is very important to us. You’ll see demonstrations downstairs of how to handle video in 3D, how we can use 3D knowledge to improve the experience in applications like videoconferencing.
Also, information retrieval, you heard Hugo on the video talk about how our interest is training information retrieval systems by using previous things that you’ve found to steer your searches in the future.
The third group, programming principles and tools, is, as I think I said earlier, leveraging a particular European competence. This group has an interest in program specification and verification, knowing software does what it says on the tin, how do you check that’s the case with software. We’ve made some very significant advances, for example, in analyzing the security properties of Web services connecting multiple machines. Is that connection of computers actually respecting the security policies of the organizations working together? Have the engineers used the right protocols to give you the security guarantees you need to protect your business? Those are questions they’re trying to answer.
We’re looking at the design of programming languages. Just yesterday I got a message from Luka’s team of a language they’ve developed helping people write device drivers, which are a very critical part of the Windows operating system, which if they’re wrong they can bring the rest of the system down. There’s still work to be done in that area, which is very challenging.
The new group I mentioned is the interactive systems group, led by Ken Wood, one of our senior researchers. This group brings a social perspective to what we do in the lab in Cambridge. The other groups are very technology, very theoretically founded. This group works the other way around. They look at how people respond to technology, how people approach technology and use that to decide what the technical opportunities are, how to deliver our technology in a form where it has more impact on society, where it’s much easier to use.
One of Ken’s colleagues, Richard Harper, will speak in the presentations this afternoon on some of the investigations he has made of what actually is going on when teenagers are text messaging. We’ve probably all got teenagers or know teenagers, we see them texting; what’s that about as a social behavior? Richard has analyzed that. It has tremendous insights as to what they’re doing and how you might organize the systems. And so I am very excited about having that in the Cambridge laboratory.
I think we’ve talked about many of our results. I can pass over these quite quickly; they’re in the statistics that Rick gave you. But we’re extremely proud of our achievements and the number of engagements and relationships that we’ve had with universities in Europe.
What I’d like to do is now just show you a couple of projects that we aren’t exhibiting. This is the list you can see downstairs; I hope you’ll find the time to see those. But just show you a couple of others that reflect some of our broader issues outside of research.
So the first one is a technology called SenseCam. You might have been wondering what this strange object is that I’m wearing on my lapel. This is effectively an automatic digital camera. It’s a prototype built in the Cambridge laboratory. It is a camera. It decides when to take a picture. This thing is stuffed full of sensors. It detects change in light, so when they turn the camera light off I can see it just took a picture. It detects motion, it detects changes in sound. It builds a record of my day.
This video playing is the researcher who built this system walking around Cambridge. With those sensors it takes just a few images when something significant happens, so you get almost like a cartoon flip book of your day. And if you’re trying to go back to what happened, the opportunity here is it’s not as dense as a video, you can see things much more clearly. You can see him walking around Kings College.
We’ve been working, for example, with a hospital in Cambridge using this device to help patients who have problems with memory. Those patients can use the input from this camera to see what they’ve done, did they remember to feed the cat, did they remember to do one thing or another, did they take their medicine and other things; extremely helpful.
We’re also discovering with this technology that it’s raising for us some very interesting social questions. How do people respond to someone walking around with a camera?
If you watched the video, you might have observed at one point it goes into a shop, then it suddenly goes blank because the researcher who was wearing it probably didn’t want to share with everyone else what she was buying in that shop and so she put the camera in her pocket. That tells us something that a device like this, it must be easy to switch it off. We wondered about making it as a necklace or a broach but it’s hard to take that off.
So those kind of questions from actually building technology, letting people explore it, that lets us understand what the social questions are and what some of the responses might be; obvious questions about privacy. This is a way to find out what people enjoy, what people dislike. Some of our researchers have taken this into a number of contexts to see how comfortable they are with carrying a SenseCam all day and my hope is at the end of the day I can download the images from this and remember all the people I spoke to, because I have a hopeless memory for names and faces. Obviously there are many ways we can go with this as an important technology.
As well as connecting the images, in fact, the camera also collects the sensor data and that gives me tremendous opportunities to learn and to mine that data. So if you want to recover, tell me what happened when I was outside, what was it that made me run away, then all that information is there.
And indeed that sensor data helps the camera. As I’m moving around, there are real challenges this camera lends to get a stable image. If you do this with a digital camera, it doesn’t take good pictures. This one does because using the sensor data it has a better understanding of what is going on, so it’s a device that’s much more aware of its context and its surroundings and I think is another exciting development going forward.
I think I’ll skip over the SenseCam plans, we’ve said these things but the hardware is just the beginning, the applications are important.
I’d like to take you to another place where our researchers go. In our computer vision group we have a lot of interest in the geometry of 3D figures, can we take two-dimensional pictures, paintings and diagrams and from those build a 3D model.
We’ve found ourselves in a fight with Mr. Hockney, a well-known artist. There is a big debate about Renaissance paintings. If you take a painting like the portrait here, this is a very early example of a realistic painting, a big transition in art when it went from being figurative to looking very representative, and so the discussion is did the artist use techniques, almost photographic techniques, camera, obscurer, projection and so forth, to capture their sketches before they painted the pictures. Some people say they do. We’ve been trying to investigate that.
If you look in that picture, there’s a mirror in the background. If we look at that mirror, there’s a chandelier hanging from the ceiling. So you might think if the artist when he was painting that picture had used projection techniques, that chandelier will be a mathematically perfect painting, and so we did the mathematics.
The original picture from the painting, we used our computational techniques to break out two of the arms of the chandelier. We then did the mathematics to put them on top of each other. If he had used a projection technique, then those mathematical projections should line up. So who thinks he used a projector, put your hands up? Who thinks he just had a good eye, put your hands up?
The people who think he painted using his eyes are indeed correct. As far as we can tell, there is no strict mathematical relationship between the two arms of the candelabra and so it’s likely he painted it as a visual image.
And so now we start another debate. When you have a painting that appears realistic, it’s not actually realistic in a mathematical sense, and so there’s an interesting discussion there about what is art, how do we interpret art, the relationship between the artistic view and the scientific view, and it’s great in research to be able to take part in those kinds of discussions.
