also available: curriculum vita (can also be used to generate list of publications) ...
research agenda
robotics
and machine learning
(see gotbots.org)
it is important that next generation computer
scientists be equipped
with the knowledge and capability to design high-availability systems
that are not just secure and reliable, but also intelligent and easy
to maintain. my current research endeavors seek to develop a deep
understanding of the complexities of intelligent robot behavior
relative to such limitations as the size and weight of the robot,
whether the robot is tethered or autonomous, battery life, the robot's
reaction time to make possibly crucial decisions regarding
ill-defined or complex situations and/or environments, and many
others. there is also a growing demand for intelligent systems. this
is not only evident through the use of autonomous mobile robots for
search and rescue missions as a result of terrorist attacks, for
example. i am interested in robotics research endeavors that focus on
cooperating intelligent agents for search and rescue, with
applications in biosurveillance, threat detection, agriculture, and
many others.
robotics: current state and
trends (march 2008; 4.4 mb; internet
connection required to access hyperlinked online resources)
intelligent classification and assessment
schemes
traditional classification techniques rely on
precisely defined
boundaries. elements in this approach are classified as belonging to
one and only one category. however, uncertainty, in the form of
fuzziness, abounds in the area of watershed classification, risk
assessment, and prioritization. the available information
characterizing watersheds under consideration can come from multiple
data sources, may have varying levels of granularity, may be
incomplete, and may also possess complex causal relationships over
space and time. in this case, uncertainty accounts for elements
possibly belonging to several categories with varying degrees of
membership. the ability to classify such ecoregions at a regional
level, and potentially extending the classification scheme to a
national level, naturally demands consideration of this inherent
fuzziness. this approach also allows the selection of boundaries, such
as classes of watersheds.
the term "soft computing" was proposed by Lotfi Zadeh
(UC Berkeley) to mean the creative fusion of neural networks, genetic
algorithms and fuzzy logic. soft computing techniques will be used to
incorporate fuzzy logic and neural network agents into a unified
classification, impairment assessment, and prioritization software
package designed to be used by regional and state watershed and water
managers and planners. the system will be designed to accept spatial
data from the user, with explicit guidance on appropriate forms of
data. these interfaces are necessary to combine various descriptions
(including spatio-temporal relations) of multiple watersheds to
facilitate the use of fuzzy measures and fuzzy integrals as
aggregation operators. these operators will be used to handle the
diverse sources of data and their corresponding features to unify the
classification and assessment schemes. the integration and use of
spatio-temporal information in a watershed classification scheme
facilitates the potential to incorporate "learning" in the
system. in this regard, neural network agents will cooperate with
existing fuzzy network agents in a multi-agent, hybrid system that
generates dynamic classification hierarchies.
educational data mining (see educationaldatamining.org)
educational institutions will continue to emphasize accountability of
all teaching and learning processes adopted. as such, assessment will
continue to be an important part of the day-to-day activities in all
levels of every academic institution. successful assessment schemes
rely heavily on data, and every educational institution gathers a
variety of data from an assortment of sources (e.g. student use of
interactive learning environments, administrative data, etc.) for a
multitude of purposes. educational data tends to be extremely complex,
spanning multiple levels of meaningful hierarchy that are inherently
dependent on the intrinsic properties of the data itself. hence,
educational data mining, or edm, is “concerned with developing
methods for exploring the unique types of data that come from
educational settings, and using those methods to better understand
students, and the settings which they learn in.”
edm can be used to facilitate various levels of university life from
student advising all the way through institution-wide programmatic and
financial planning.
more information for prospective graduate students ... next
research projects
institute for research in intelligent systems
(IRIS)
- check out the IRIS website at
iris.ecst.csuchico.edu
...
intelligent systems laboratory
(ISL)
- check out the ISL website at
www.gotbots.org
...
chico
statements article about the isl ...
