Data Fusion (MIT-LL)
After completing my bachelors degree, I worked for two years as an algorithm development engineer in the Biodefense Systems group at MIT Lincoln Laboratory, a federally funded research and development center (FFRDC) operated by the Massachusetts Institute of Technology.
At Lincoln Laboratory, I was a member of two project teams. My primary project involved the development of information fusion algorithms for biological standoff sensors. The goal of this project was to develop algorithms and decision support tools to improve the detection of biological weapons attacks from sensor data. From a mathematical standpoint, we were concerned with the detection of rare events using multiple data sources. Since biological attacks occur infrequently, the sensors used to detect these events alarm only occasionally, and when they do, it usually represents a false alarm. False alarms can be expensive in terms of time, productivity, and resources; however the cost of failing to detect a real attack is even greater.
Our task was to use the information from one of more biological sensors and other available information sources to determine which sensor alarms are true alarms and which are false. Our algorithms combined data from multiple biosensors, meteorological data, and a simple transport and diffusion model for aerosol plumes to improve detection capabilities while significantly reducing the number of false alarms. Our team developed two algorithms for the task, one that quantifies information corroboration, and another that uses supervised machine-learning techniques. We also extended our algorithms for threat mapping and predictive modeling.
I also worked on a side project developing artificial intelligence algorithms that model cognitive processes as part of the lab's Cognitive Robotics initiative.
Please visit my Publications page for additional information about this work.