​​​​Robert E. Marmelstein, PhD

Worked with a grad student to develop a cloud-based testing service for the client-side of mobile applications.  The idea is to be able to simulate many clients (hundreds to thousands) using the app in order to stress test the server.  We hosted the server and the testing suite on Amazon Web Services.  The target app for this research was a Blackjack program.  [Student:  Paul Mirams]

This is a relatively new research effort to improve upon the Artificial Intelligence (AI) player that comes with the Open Red Alert (ORA) game.  This research has two components.  The first is to create a series of tunable rules for the ORA-AI.  The second is to make use of a genetic program to tune these rules using a relatively large population (500) over at least 100 generations.  The goal is to come up with a robust ORA AI player that can beat the current HackyAI implementation, as well as, skilled human players.  [Students:  Josh Soberg, Jeff Greener]

This work involves the evolution of genetic program to trade stocks using certain time-sensitive strategies.  To date, two strategies have been successfully developed.  The first chooses stocks to buy prior to the predicted dividend declaration date.  The second choses stocks which have just undergone a correction in price of 10% or more.  The idea in both cases is to pick stocks that will rapidly rise in value in a short period of time (60 days or less).  Successful genetic programs were evolved for both these strategies using real stock market data from the 2001-2015 time period.  Under publications, three (3) papers discuss this research.

[Students:  Andrew Yatsko, Alexander Hunt, Christopher Eroh]

The BioMiner system enables users to perform disease or symptom surveillance using real time HL-7 (demographics, pharmacy transactions, ICD codes, lab/test results, chief complaints, medical observations, etc) transaction streams from medical facilities. In addition, BioMiner monitors symptoms and trends across your patient population, producing statistical summary reports of its findings.  The system employs expert knowledge to automatically detect rare, but lethal diseases (such as inhalation anthrax) and instantly alert medical professionals to the situation. It also provides an alert based system to notify personnel to the existence of these patients and provides any observed evidence regarding their current situation. 

[Student:  Marc Kurtz]