The focus of my research has changed over the last six years.  My earlier research focused on developing techniques for the construction of robust, hand-crafted expert systems.  My current research interests are in the area of data mining and knowledge discovery where the goal is to develop methods and techniques for automatically acquiring new and useful knowledge from large databases.  I currently have three active research projects in this area.  The first two projects deal with large quantities of remotely sensed scientific data.  The third project deals with extracting useful information from large quantities of computer network audit data.

Knowledge discovery from oceanographic data.
For the past four years, Julia Hodges and I have been working with scientists at the Naval Oceanographic Office (NAVOCEANO) on a project in which we are developing techniques for knowledge discovery from oceanographic data.  Scientists currently spend an enormous amount of time analyzing sonar images manually in order to identify “provinces” of the ocean floor.  The process is tedious and error-prone.  We are developing data mining algorithms and a knowledge discovery system to help automate this process.  This research was initially funded via a DEPSCoR grant from ONR for three years; subsequent funding has been directly from NAVOCEANO.  A journal article and a number of conference papers have been published dealing with this work.
OKEANOS

Knowledge discovery from remotely sensed agricultural data.
This project was initiated jointly with Dr. Jac Varco from Plant and Soil Sciences and Dr. Marvin Salin from Biochemistry in 1999.  In a project sponsored by NASA through the Remote Sensing Technologies Center, we are developing algorithms to determine the nutritional status of crops from remotely sensed data.  One student has completed a Master’s degree in work related to this project and several papers describing this work are in preparation.  We have applied for continuing funding for the next two years.

Data mining applied to intrusion detection. 
Dr. Rayford Vaughn and I direct a joint project in which fuzzy data mining techniques are used to detect intrusions from audit data collected from computer networks.  A journal article and several conference papers have been published that describe the work.  Additional papers are in preparation.  Seed funding for support of two graduate students for this work has been obtained from the VP for Research, the College of Engineering, and TVA.  We have been notified that the Army Research Laboratory will fund the project for one year.  Proposals for additional support have been submitted to several government agencies.  Dr. Anthony Skjellum has joined us as a co-investigator on several of these proposals as we have expanded the scope of the  research area to address issues in the cluster computing environment.