Border Security: Supplementing Human Intelligence in a Sensor Network Using Sequential Pattern Mining

Sensors are being deployed to improve border security generating enormous collections of data and databases. Unfortunately these sensors can respond to a variety of stimuli, sometimes reacting to meaningful events and sometimes triggered by random events which are considered false alarms. The intent of this project is to supplement human intelligence in a sensor network framework that can assist in filtering and real-time decision making from the large volume of data generated. Our conceptual design of a human-computer system is to use off-line learning to identify the important patterns. The critical real-time system uses the identified patterns from off-line learning in a system that relates the risks of false alarms with the length of patterns and the time interval distributions between sensors in the patterns to allow the human to generate intervention decisions. The human would supplement the computer information with the current threat levels and the available resources for reactions.