Noblis has three additional internships open as part of our 2008 Summer Internship Program for undergraduate students. Our goal is to provide the students with a rich and challenging work experience, foster a positive relationship with Noblis researchers, and nurture a passion for our client missions and company. Noblis has been a part of the CMPS Corporate Scholars Program for three years and a proud supporter of Maryland academics.
The program will run for eight weeks from Monday, 9 June, through Friday, 1 August. Intern assignments are located in Falls Church, right off the Beltway.

The program is further described at http://www.noblis.org/SummerInternships.htm. Interested students should contact Roger Snelling at roger.snelling@noblis.org or at (703) 610-1688. Please submit a cover letter and resume to Mr. Snelling for each position.

Salary range: $15-$17

Remaining Internship Assignments:

Eigenimage-based Face Recognition System
Background: The backlog of passport applications has made our work on streamlining the passport process at the Department of State a high profile project. The intern will assist in the development and implementation of image processing and machine vision algorithms for biometric identification. In particular, the intern will assist with the research and development of an eigenimage-based face recognition system which would be used in our study on one-to-one face recognition for the Department of State. At the end of the summer, the intern will have contributed to the project by developing her own algorithms for image enhancement and will have the opportunity to present her work at the Department of State.

Qualifications: Candidate may be an undergraduate in mathematics, statistics, or computer science with an interest in image processing, pattern recognition, and machine learning. This project will include some software development.

Applying Machine Learning Algorithms to 1:N Data to Support Customs and Border Protection Targeting

Background: The intern will research classification approaches and machine learning algorithms to identify ways to apply machine learning effectively to data sets that have a variable number of attributes describing each item to be classified. Machine learning algorithms typically classify items based on a feature vector of fixed length. The machine learning algorithm then tries to find the characteristics that best predict which items will be of interest, learning from a training set of items whose classifications are known. Noblis clients are interested in applying machine learning techniques to targeting for high risk shipments. One challenge is that data is not natively in flat file format. Instead, many aspects involve one-to-many and sometimes many-to-many relationships. For example, a single shipment identifier may have several ports associated with it, several commodities, and several scores at different times. A single physical shipment may have multiple shipment identifiers. A single shipment ID may be associated with several containers, or several shipment IDs may be associated with a single container. The goal of the intern’s research will be to investigate if there are particular machine learning algorithms that are better suited to handle one-to-many data, or strategies that humans can take to convert the one-to-many data to one-to-one data in ways that will be most useful.

Qualifications: Candidate must be an undergraduate student in mathematics or computer science, engineering with an interest in machine learning algorithms and with an ability to apply knowledge gained to problem at hand.

Secure Freight Initiative (SFI) Discrete-event Simulation
Background: The Intern will design, develop, and demonstrate the secure freight initiative discrete-event simulation model, with animation, depicting:
• geographical placement of equipment at a major foreign ocean port
• scanning and associated operations on cargo containers bound for the US
• resulting impact on time-through-the-system, throughput, queue lengths, potential bottlenecks, etc
• have a user input component to facilitate changing model parameters

• The discrete-event simulation model will be developed in Arena or another appropriate application.

Qualifications: Candidate must be an undergraduate student in operations research, applied mathematics, or computer science with an interest in modeling and simulation.