DIMACS/CCICADA Student Run Seminar, Fall 2009

This student-organized seminar series is open to all, and will strive to invite speakers in the mathematical, statistical, and computational sciences whose work can be applied to problems related to homeland security. The Department of Homeland Security is a wide ranging agency with many types of problems including: detection of nuclear material, sensor location, evacuation routing, speech recognition, data analysis and exploration, risk analysis, biological outbreak detection and control, and visual analytics. Of particular interest will be risk analysis, speech and visual pattern recognition, and new methods of data analysis and exploration.

All talks are in CoRE 431, the 4th floor DIMACS seminar room unless otherwise stated.
Next Seminar:

Date: October 29, 2009
Time: 12:00 PM
Speaker: Paul Raff, DIMACS Postdoc
Title: Stemming the Flow of Information
Abstract: Analysts in the government (and anywhere else) state a problem that only gets worse: the glut of information. The amount of information that needs to be analyzed far exceeds the manpower that is available. I will talk about the large scale program, blackbook, being developed by the government, and BOXER, a document classifier suite that works on its own or within blackbook to help the analyst. In this talk I will cover the fundamentals of online Bayesian regression, the cornerstone of BOXER, and will show how BOXER interacts with blackbook in a mock scenario.

BOXER is being developed under the MMS (Monitoring Message Streams) program in DIMACS, with Paul Kantor as the research head. Research and Development supported by the KDD (Knowledge Discovery and Dissemination) program of the Intelligence Community.

Upcoming Talks:

Date:
Time: 12:00 PM
Speaker:
Title: TBA
Abstract: TBA
For a list of previous talks see the DIMACS/CCICADA seminar archive.
If you are interested in giving a seminar talk, send an e-mail to or otherwise inform Emilie Hogan (eahogan at math dot rutgers dot edu).