This page is no longer being updated and remains online for informational and historical purposes only. The information is accurate as of the last page update.
For questions about page contents, contact the Communications Division.
Like an early prospector, Trustee Scholar Bryan Culbertson ’07 (Charlottesville, Va.) headed west this past summer and struck gold. The computer science major served a coveted internship with Microsoft Live Labs in Redmond, Wash.
A relatively new applied research lab that has close ties to Microsoft Research, Microsoft Live Labs is part of the industry research arm of Microsoft Corporation. Culbertson and about 10 other interns spent 12 weeks working in the Machine Learning and Applied Statistics Group with 20 Microsoft researchers.
“While I was there, I researched social gaming,” he explains. “My mentor, Max Chickering, and I worked on what Louis von Ahn of Carnegie Mellon University called ‘harnessing human cycles’ in computation. The idea is that there are many things that humans do better than computers, such as recognizing images. The difference between a tulip and a rose, for example, can be identified by most five-year-olds. Computers, meanwhile, still have problems identifying the difference. What we wanted to do was capture a human’s capability of doing this through automated processes.”
The challenge is keeping users engaged while providing useful data. Culbertson and Chickering, a principal scientist, developed different techniques to entertain users so they still supplied useful data. For instance, if Culbertson and Chickering merely displayed pictures for users to view one after another and asked what the pictures were, the repetitive nature of the exercise would become boring and the users would lose focus.
“Instead, we came up with different entertainment styles and games to keep them entertained, while they labeled the images,” Culbertson says. “The game was critical to the process of gaining the information we wanted. The goal was to come up with problem types that are good for the human computation and techniques that keep people entertained.”
Culbertson and Chickering determined it was necessary to provide an incentive for participants, so they offered social rewards as a way of entertaining.
“It’s currently in line for a patent,” Culbertson says. “The idea is that players have partners in this game, and they are anonymous at the beginning. As they play the game, information about one another is revealed to the partners. Maybe their images are blurred and come into focus, or reveal what state they live in.”
Culbertson and Chickering developed a framework for further development of this incentive structure. To create the framework for testing, Culbertson learned a set of web application development technologies called ASP.NET. and was exposed to proper test procedures. The framework Culbertson helped develop is currently in use at the lab.
Internships with the corporation are highly sought after and difficult to obtain. It is especially difficult for undergraduates to be selected. Chickering points out that of the 243 interns at Microsoft’s labs this summer, more than 92 percent of them were Ph.D. students.
“[Microsoft] is one of the most difficult internships to get, along with opportunities at Google,” notes Chun Wai Liew, associate professor and head of computer science. “There is a rigorous screening process before the students are selected.”
At the time he took the assignment, Culbertson still was debating whether he wanted to take a “regular” job or go into industry or academic research.
“I was considering all options,” he says. “Taking this job and being able to contrast it with the interns that were working for Microsoft Corporation, I recognize that research is definitely the area that I want to be in. It’s a fulfilling extension of the academic research I’m doing at Lafayette, but it’s necessary for me to go to graduate school first. My experience at [Microsoft] has cemented my desire to go to graduate school.”
Culbertson also took the internship to explore a “new area of the world.”
“I found out that Seattle is a place that I enjoy, and my experience has influenced my decision to apply to the graduate school at the University of Washington, among others,” he says.
He also felt comfortable in the corporate culture at Microsoft, which included casual dress, regular meetings with Chickering while playing frisbee, and many professional development opportunities. He attended daily seminars presented by professors, industry experts, and people at Microsoft who give talks about interesting topics ranging from their own personal research to issues outside the field.
“Microsoft is trying to provide an academic learning community in addition to the actual work you’re doing,” Culbertson explains. “You can attend these seminars as often as you want if you have the time. I learned so much about the computer science field. Researchers came in from Washington State, Berkeley, and Carnegie Mellon. Some were out of my direct field of research, but the experience was very enlightening and very rewarding.”
Liew, who collaborated with Culbertson on EXCEL Scholars research to develop models of how fish swim, believes the internship experience will have a tremendous effect on the student as he continues his academic career.
“The researchers have different perspectives and interests,” Liew explains. “It will help Bryan in getting into a top graduate school, give him a better idea of the different research areas in computer science and artificial intelligence, and help him to select research areas and topics for his Ph.D. work.”
Chickering agrees that Culbertson has a bright future in the field.
“Bryan really impressed me this summer with his creativity, his insight, and the enormous amount of work he got done,” he says. “After working with him, I have no doubt that he will be a superstar in graduate school.”
For now, Culbertson is excited to bring the knowledge he gained back to campus, where he is currently working with Liew on honors thesis research to examine ways in which resources can be automatically allocated to complete a set of tasks through dynamic negotiation among a collection of peers.
“He is working on a very tough problem,” Liew explains. “Students need to participate in the research process so that they can be better prepared for graduate school. It also gives faculty a better opportunity to evaluate them for their research skills and potential, both of which are vital components of graduate school recommendations. If students can generate a publication or presentation from their research, [it] would enhance their profile.”
Culbertson’s EXCEL work with Liew resulted in presentations at the annual Society for Integrative and Comparative Biology conference in Orlando, Fla., and the 20th annual National Conference on Undergraduate Research at University of North Carolina-Asheville.
Culbertson was also awarded a $1,000 Upsilon Pi Epsilon scholarship for this school year because of his class work and research. Upsilon Pi Epsilon, the national honor society for computer science, awards only 10 scholarships per year; in the last five years, all four students who have been nominated by Lafayette have received scholarships.
A past recipient of the James P. Schwar Prize, Culbertson is a member of the Reeder Scholars program, Kayaking Club, and Association for Computing Machinery.
Selected from among Lafayette’s top applicants, Trustee Scholars like Culbertson have distinguished themselves through exceptional academic achievement in high school. Lafayette provides them with an annual minimum scholarship of $7,500 ($8,000 effective with the Class of 2009) or a grant in the full amount of their demonstrated need if the need is more than $7,500.