|
My discipline is computer science.
This field is large and relatively new amongst the sciences.
Within this field, the subfield that interests me most is Artificial
Intelligence (AI).
This area has suggested having strong promises since its
infancy in the 1950s. However, in present day we are finally able to
its usefulness actualize: from autonomous vehicle navigation, a myriad
of recognition systems (such as speech, face, hand-writing, eye, etc),
anomaly detection, game playing (such as chess system Deep Blue),
classification schemes, and even prediction systems related to stock
markets or which music CD we will likely be interested in buying.
Artificial Intelligence is primarily concerned with the ability and
task to make computers think abstractly--to have intelligence. For
example, can we develop a system that never loses against a human
player in a game of tic-tac-toe? Sure, AI could be used; we could
program a computer to KNOW all the rules of tic-tac-toe ahead of time
and to appropriately move such that it will never lose. However, a
system that has the ability to learn
from its past behavior possesses a higher level of abstraction. This
sub/related field of AI is called machine
learning and interests me most.
Within machine learning, the
areas and problems about which I am most curious are planning, decision-making within uncertain
environments, reinforcement learning, and information retrieval.
Dr. Philip Chan was my first research advisor,
and
throughout my three years of undergraduate research he served as a good
mentor who demonstrated to me the nature and way to conduct sound
research. My past projects are outlined below:
Adaptive
Web Personalization:
For sixteen months I have researched adaptive web personalization.
Specifically, my research advisor and I have been interested in finding intrinsic,
non-invasive ways to model a web user's behavior so as to accurately predict what
she is interested in. Using only a web user's bookmarked web pages, I developed a
system that evaluates and ranks web pages comparably well to that of Google's
ranking. Moreover, with the useful information of a given Google ranking, and
appropriately combining it with our respective ranking, our system outperforms that
of Google's ranking alone.
Speech
Recognition:
protected
NASA Anomaly Detection:
During
the summer after my sophomore year, I was introduced to artificial
intelligence while serving as a researcher on a NASA-funded
anomaly detection project. As the only undergraduate on the four-person
team, I was responsible for exploring ways to dynamically model data
that are to be considered "normal" This sixteen-month-long project
provided me with
both an understanding of how research is conducted and with the
technical knowledge that included clustering, statistics, distance
measurements, and ROC curves. Ultimately, I contributed toward
providing NASA with my team's patent-pending product.
|