Abstract
The k-means algorithm is used to cluster data using an iterative process that repeatedly searches for the nearest neighbors to a set of cluster centers. It can be parallelized with the map and reduce patterns. Implementations are given here in Intel Cilk Plus and Intel Threading Building Blocks (TBB), along with fusion optimizations, custom hyperobjects (in Cilk Plus), and use of thread-local storage (in TBB).
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have