Classroom contest yields publishable results
In the 21st century, design contests have emerged as a way to make rapid progress on tough computational problems. The million-dollar Netflix Prize, which sought to improve Netflix’s movie recommendation algorithm, is probably the most high-profile example. But similar, if lower-stakes, contests have addressed problems in computer vision, medical-data analysis, and weather prediction. In 2012, two PhD students in the lab of Hari Balakrishnan, the Fujitsu Professor of Computer Science and Engineering, hatched the idea of bringing that type of contest into the classroom. The next semester — spring 2013 — the graduate-level networking course that Balakrishnan taught featured a two-week research project in which 20 two-person student teams designed competing protocols for managing congestion in cellular networks. That same spring, Balakrishnan’s research group was scheduled to present such a protocol — dubbed Sprout — at a major networking conference. So the prize for any team that could better Sprout’s performance was...