Mathematics Colloquia and Seminars
Exact recovery of clusters via k-means SDP relaxationStudent-Run Applied & Math Seminar
|Speaker:||Yang Li, UC Davis Mathematics|
|Start time:||Wed, Nov 16 2016, 12:10PM|
The classic Lloyd's algorithm solves the k-means problem efficiently but has no theoretical guarantee. However, by considering an SDP relaxation of the k-means problem, we can derive exact recovery under the random ball model, and the question becomes what is the necessary separation between these balls.
RSVP for Pizza