| The world is awash in data we cannot
understand or use. A homegrown technology is now making sense of
this seemingly random knowledge to detect computer hackers, personalize
television commercials, or locate a missing ship at sea.
Canada's growing expertise in applied mathematics could have
major implications for national defence, network security, financial
markets, and search and rescue. It may even alter the face of television.
Dr. Michael Kouritzin is a world-renowned expert in mathematical
and statistical sciences at the University of Alberta. He is also
the president of two new companies spun off from the university
and MITACS that are bringing his novel statistical algorithms to
large corporations like Lockheed Martin.
|
MITACS-developed technology is helping an Edmonton
company track performers around a stage
Acoustic positioning research is combining robotics
and mathematical filtering theories developed at the
Prediction in Interacting Systems (PINTS) laboratory
at the University of Alberta to design a virtual stagehand
that keeps lights shined on a performer based on predictions
of where the person will move.
A computer receives ultrasonic signals from a tracking
device worn by a performer to track and control the
stage lighting. |
|
Dr. Kouritzin heads the Prediction in Interacting Systems (PINTS)
– a MITACS-funded research centre formed in 1999 that is marrying
mathematics, statistics, and computer science to determine what
is going on in real life. PINTS has developed mathematical modelling
and prediction techniques that take into account all the behaviour,
characteristics, and conditions that affect how random systems operate
and behave. It removes the guesswork and human error when it comes
to tracking boats, aircraft, submarines, and other targets.
"Imagine a dinghy lost at sea," explains Dr. Kouritzin.
"A helicopter is trying to locate it based solely on noisy
imagery taken from a digital camera, infrared, or radar. Our platform
filters through this data and comes up with a statistical prediction
on where the dinghy is located. This same mathematical filtering
theory can be used to track computer hacking, money laundering,
or even insider trading."
Lockheed Martin, the main corporate partner in PINTS, has filed
three patents from the technology, two of which are being used in
its current research.
"We're defence contractors, so we deal with military applications
– trying to find the locations, speeds, velocities of non-cooperating
platforms, such as an unfriendly jet air fighter, or missile,"
says Dr. Ron Mahler, a staff scientist at Lockheed Martin Naval
Surveillance Systems in St. Paul, Minnesota. "To do this research,
we need the type of technology that Mike has been developing."
Dr. Mahler describes the Canadian researcher as a pioneer in an
area of applied research that has stymied engineers, mathematicians,
and computer scientists for years.
"These technologies are definitely ahead of the game, so much
so that our sponsorship of his work appears to have spurred a huge
increase in interest in non-linear filtering," he adds. "Mike
presented his first paper on this in 2000 (at the AeroSense-2000
Conference in Orlando, Florida), and since then, the number of papers
in this technology area has increased exponentially."
Dr. Kouritzin is now actively marketing the technology through
two spinoff companies, Fast Track Technologies and Random Knowledge.
The first products are in the areas of network security, fraud detection,
finance, and television.
One product, called DEFEND, is an intrusion detection and response
system that finds, isolates, and destroys network intruders such
as spyware. Random Knowledge is currently seeking venture capital
funding to finalize development of a prototype.
"We've used the PINTS technology
for some small contracts, and as part of our ongoing
research. We're hopeful the technology will help Lockheed
Martin win large contracts in the future."
Dr. Ron Mahler
Staff Scientist
Lockheed Martin Naval Surveillance Systems |
|
Fast Track and Random Knowledge are also working with Invidi Technologies
Corp. of Edmonton to develop a product that would use PINTS algorithms
to detect and track the optimal viewing audience for specific commercials.
"The idea is to profile television viewers, so the right commercials
can be played. Depending on your viewership profile, you might get
a different commercial than your neighbour." Field tests are
planned for early next year, with a commercial launch expected later
in 2005.
One of the greatest challenges facing mathematics today is moving
it out of the universities and into real-world applications. Dr. Kouritzin
credits MITACS for making significant strides in this area.
"MITACS is playing an important role in getting the academic
community to realize and exploit the industrial relevance of mathematics,
and getting industry to realize how mathematics can be used to improve
efficiencies, productivity, and profits. What we're doing now with
Random Knowledge is just the beginning."
Simulations of the technology can be found on the PINTS Web site
at www.math.ualberta.ca/pints.
The Random Knowledge Platform
www.mitacs.ca
|
