Projects
-
Development of Novel Algorithms for Analysis of Biosequences
With the recent developments in high-throughput
sequencing, sequencing entire genomes is no longer cost prohibitive. Analyzing the massive amount of data
produced through these methods requires the development of novel tools that can detect features of
interest between different sets. The goal of this project is to design and develop tools to identify and
classify features within these datasets.
-
Discovery of Action Rules
The goal of this project is to design and implement
algorithms to discover action rules from a collection of data. An action rule represents a series of changes,
or actions, which can be made to some of the flexible characteristics of an information system that ultimately triggers
a change in the targeted attribute.
Diagnosis and Grading of Alzheimer's Disease
The goal of this project is to develop machine learning
technology to accurately predict the progress of Alzheimer's disease.
-
Multi-Layered Vector Spaces (MLVS) Model
The goal of this project is to construct a mathematical habitat
for allowing
and accommodating mathematical models capable of comparing, classifying and analyzing various features
of biological sequences, both at the micro and macro level.
-
NSF CISE / SBE Discovery in a Research Portfolio
This project resulted in the development of a Proposal Information Management System
referred to as the Concept Map-based Organizer for Research Portfolios (C-MORE).
The C-MORE system features both top-down analytic navigation and query-based information exploration.
It is able to provide decision support at both managerial and strategic levels.
-
Algorithms for Discovering Potentially Interesting
Patterns
A pattern discovered from a collection of data is usually considered potentially
interesting if its information content can assist the user in their decision making process. To that
end, we have defined the concept of potential interestingness of a pattern based on whether it provides
statistical knowledge that is able to affect one's belief system. Three algorithms have been designed
and implemented to discover such patterns: Discovery of All Potentially Interesting Patterns (DAPIP),
All-Confidence based Discovery of Potentially Interesting Patterns (ACDPIP), and ACDPIP-Closed.
-
Knowledge Hiding in Databases (KHD)
We have developed a methodology to analyze the impact of data mining technology
on database security. Our methodology consists of six steps that make up a new data analysis process
called Knowledge Hiding in Databases (KHD). The goal of the project is to design algorithms to support
the non-trivial hiding of knowledge. We define the term non-trivial to imply
that knowledge is concealed in a manner that maximizes the amount of
released data and also maintains to the greatest extent possible the
integrity of the data.