My research areas are data mining and reinforcement learning. I'm a member of the Auton Lab which is a data mining lab headed by Andrew Moore. Below are several projects that I'm currently working on and that I have worked on in the past.
I am currently working on the early detection of disease outbreaks. My main research project is WSARE (What's Strange About Recent Events), a non-specific early disease outbreak detection system. WSARE works by comparing recent events against a baseline and reports what the strangest group of events that it found for the recent period. More information can be obtained at the Auton lab. If you are looking for data that was used in the WSARE v3.0 paper, check out this link
My work on Disease Outbreak detection requires the use of Bayesian networks. Andrew Moore and I have worked on a fast structure learning algorithm called Optimal Reinsertion.
The Cuevas algorithm is a technique for estimating the number of clusters in a dataset. The algorithm can also be used for finding clusters against a noisy background. However, Cuevas' algorithm is too computationally expensive. We have implemented a modified version of Cuevas' algorithm that addresses the computational problems for lower dimensional but dense datasets (as opposed to text data which is high dimensional but sparse).
I spent the summer of 2000 working with the Information Economies Project at IBM in Hawthorne, NY. We had lots of fun creating bidding strategies for agents involved in a continuous double auction. I've also worked on an algorithm for finding epsilon Nash Equilibria for games with continuous payoff functions although this work is currently unpublished.
Jeff Schneider, Andrew Moore and I have worked on a distributed reinforcement learning algorithm which uses distributed value functions. This work was applied to a power grid simulator where power distributors decided how power from power plants were allocated to cities.