Summer Interns Dive Into Solving Algorithmic and AI Problems
From studying quantum simulation and bias in artificial intelligence (AI), to advancing post-quantum cryptography systems and machine learning techniques, 22 undergraduates from across the country are wrapping up their summer research internships at the University of Maryland.
The students are part of the Research Experience For Undergraduates (REU) program to study Combinatorics, Algorithms, and AI for Real Problems (CAAR). The prestigious 10-week program is designed to bridge the gap between the theories on algorithms, probability and combinatorics, with other computer science disciplines like AI, operations research, and practical machine learning. This year, CAAR is being supported by the National Science Foundation (NSF), Google, and UMD alum Brendan Iribe.
In addition to gaining research experience and working with expert faculty, students are reimbursed for room, travel and food, in addition to their $6K stipend. 
“This year's group of students are outstanding,” says John Dickerson, an associate professor of computer science and one of the program’s leading principal investigators (pictured far right). “They’re technically top tier and creative, with a real social focus.”
Dickerson is leading CAAR with William Gasarch, a professor of computer science (second from left). The competitive program had nearly 100 applicants this year, says Gasarch, who has been administering it since 2013. There are a total of seven projects to participate in, two of which are directly overseen by Dickerson and Gasarch.
“Many of the projects also involve grad students, and some of them involve high school students,” says Gasarch. “Hence, they are intergenerational as well as interdisciplinary.”
Research intern Abdulaziz Memesh describes Gasarch as “a funny guy with infinite math shirts,” and explains how he and his assistant Auguste Gezalyan went out of their way to create a friendly environment for his cohort. “I don't think of the other students as colleagues, but as a family,” he says.
Memesh, a junior studying computer science at Georgia Tech, adds that this experience has taught him invaluable lessons about what conducting research is really like, including how to take responsibility and be proactive.
Justin Huang, a senior at Pennsylvania State University double majoring in computer science and statistics, has been working with Assistant Professor Furong Huang to reduce bias in machine learning applications.
His favorite application has been analyzing the demographic data of bank loan applicants.
“There’s a history of denying loans to people of color,” he says. “So how do you decide who will default on a loan when you know that the data is historically skewed? We’re looking at how to avoid repeating these mistakes and get away from these biases.”
Yang Hong, a senior and mathematics and computer science double major at Bucknell University, says that this is the first time he’s worked with such a big research team, and it’s given him a new appreciation and deeper understanding of group cooperation.
Dalal Ahmidouch, a junior at Wake Forest University who is also double majoring in computer science and mathematics, has been working to make images unrecognizable using machine learning techniques like perturbations.
“The goal is to improve privacy so that other people don’t scrape their images and use them without the owner’s consent,” she explains.
That project is being overseen by Tom Goldstein, an associate professor of computer science and the new interim director of the University of Maryland Center for Machine Learning. Goldstein, Dickerson and Huang all have dual appointments in the University of Maryland Institute for Advanced Computer Studies (UMIACS).
The rest of the projects are being led by faculty who also have dual appointments in UMIACS and the Department of Computer Science: Professor Andrew Childs, Assistant Professor Ian Miers, and Professor David Mount; in addition to Associate Professor Dana Dachman-Soled who has a dual appointment in UMIACS and the Department of Electrical and Computer Engineering.
—Story by Maria Herd