Graph matching via convex relaxations

Graduate Student Seminars
Dec 8, 2015
12:30 pm
Fine Hall 214

I'll talk about a few convex relaxations for graph matching, encoded as a quadratic assignment problem (QAP). I'll discuss a popular, but expensive, semidefinite programming (SDP) relaxation for the QAP by Zhao et al. If there's time, I'll talk about how we might relax it further to solve much larger problems in the case where one of the graphs is sparse.