Imagine being able to explore heavenly bodies in the sky and discover new
worlds just like professional astronomers do, but without having any knowledge
of astronomy. Or how about searching the Internet together with your colleagues,
friends, or family members, but from different desktops sitting anywhere in the
world? How about instantly search for any object in a video without having to
run it manually?
Those were just three of the 35+ projects that were showcased at Microsoft's
TechFest 2008, the annual event organized by Microsoft Research.
“This is the only time of the year when we get all of our research labs to
show their developments”, said Rick Rashid, Senior VP of Microsoft Research
during the keynote. “We've grown to 800 PHD researchers around the world”, he
added. Plus, there are hundreds of visiting faculties and interns at these labs,
and 15% of MSR's budgets are invested in universities for labs grants, research
grants, fellowship programs, etc. Over 3,700 peer reviewed papers have been
published by these labs so far.
While MSR's activities do help foster a healthy research environment in
academia around the world, they also help transfer new technologies into
Microsoft products themselves. Just about every MS product has benefited from
this research, whether it's Vista that got its sidebar or Excel 2007, which uses
smart chart labels. There's a considerable amount of technology transfer that
happens from MSR into MS products.
Besides technology transfers, MSR labs are also researching on how IT can
resolve some of the world's most pressing concerns, like discovering an AIDS
vaccine, combating epidemics, enhancing socio-economic development, etc.
Project Demos
World Wide Telescope: A completely new way of looking at the sky. It's
basically a rich visualization environment, a virtual telescope created with
data gathered from the world's biggest terrestrial and space telescopes. A user
can pan and zoom across the night sky to identify constellations, galaxies,
black holes, nebulae, etc. It's so easy to use that even a child can use it. The
desktop version of this virtual telescope is expected to be made available for
download soon.
Collaborative Internet Search: Today, most Internet search tools are
meant for a single person only. This becomes a real challenge when you need to
collaborate. What if you're working on a business project with your colleagues
and you need to search for something together? The solution today is to either
use separate machines for the job, or fight it out on a single machine. Separate
machines duplicate efforts, while single machine makes it difficult to work
together because everyone wants to follow their own approach. So there are two
approaches that have been created to resolve both of these issues. One is called
Search Together that allows each collaborator to use his or her own computer,
and the other is CoSearch where multiple people can collaborate on the same
machine.
Singularity: A prototype OS for computer science researchers. Its
Research Development Kit is available for download from CodePlex, Microsoft's
open source project hosting website. We've also given it on this month's DVD,
and would be covering it in a later issue.
Interview with P. Anandan, MD, Microsoft Research India |
The discussion with Anandan
covered a wide range of topics, from algorithms to socio economic
development projects; providing some very useful insights into the exciting
field of computer science research. Here are a few excerpts from the
discussion.
Algorithms are the latest
research group in MSR India. Please tell us about this field.
We have 7 research groups at MSR India as of
now. Last summer, Dr. Ravi Kannan, a Sr. Algorithms professor at Yale
University joined us. An algorithm is an interesting area that has
communities of Indian researchers working both within India and in the US.
As a company, MS is moving into a lot of new areas, like search, wireless
networking, etc that have some very interesting research problems, at the
core of which are complex algorithms. Another reason for having an
algorithms group is that they're undergoing transition. They used to be
theoretically focused a few decades ago. Their researchers would hardly
write any code, and were busy proving the correctness of algorithms,
analyzing their complexity, etc. But as a community, they realized that
there are big problems-billions of nodes in routing networks, billions of
entries in databases, literally billions of web pages. In many of these
cases, the data is not stored on a single computer. It's not available from
a single source, and storing all the data and then processing it is
virtually impossible. So, one has to rely on streaming algorithms, where
data is processed as it comes in. There are also statistical algorithms,
which use probabilistic methods to do things, giving probabilistic/
statistical answers instead of definite ones. That's a new and exciting
research direction for the field itself.
India has many fields that require
considerable research, e.g. power and utility sector, wherein data centers
are coping with the problem of power consumption. What research is happening
in this area?
In a certain sense, societal problems provide ideas and opportunities
for you to explore. We've pursued many problems that other labs elsewhere
would not have seen. Once you think of the problem, you'll think of its
application in a different context for elsewhere in the world. Even if the
problem is localized, the solution or technology that you develop almost
always applies to every place in the world for many different kinds of
problems that even you may not have thought of in the first place. Take our
Robust Locations research for instance, which provides a new way of looking
up an address in an online map. Existing maps like Google maps presume a
certain format to your address, e.g. House number, street, city. They look
at that structure and parse it. But the addresses in India are very
different, e.g. in Bangalore, a typical address would be like 7th main and
6th cross, Sadashiv Nagar, Opposite RTO office. It has a bunch of queues and
tokens, landmarks etc. Plus, it can be in so many different languages as
well. So we developed an address geo-coder that doesn't assume any
particular format. It works really well, and we've been pursuing it with the
product groups at Microsoft. It works even when there are spelling errors.
Finally, the addressing was a problem we thought of due to the environment
we're in, but its solution could apply to any other country.
Is there any research related to social
networking?
We have a social networking group, which is developing technologies for
emerging markets. We did a project that we called Combine, which was
inspired by the fact that in India, usage of cellphones for information
sharing info is much more common than anywhere else. If people want to
download a media file such as an image, audio, video, etc, it's very slow on
the mobile phone. We've created a system that will automatically split the
media source into smaller pieces and have several cellphones in the
neighborhood download them. These would then all be send to one phone via
Bluetooth and combined. There are challenges in this, e.g. we don't know how
many cellphones are around, whether somebody would allow his/her cellphone
to be used for this job, security and privacy issues, etc. But ultimately,
the idea here is to use a mobile device for a collaborative and collective
effort. |
Anil Chopra was hosted at Redmond by Microsoft Research