yakmo

v 2015-01-21

C++ implementation of robust, efficient alternative k-means clustering

Yakmo implements robust, efficient k-means clustering with triangular inequality and smart initialization, while supporting alternative clustering outputs. The use of the triangular inequality allows k-means to skip unnecessary distance calculations, while the smart initialization by randomized seeding (k-means++) not only improves solution accuracy but also accelerates the algorithm convergence. In addition, you can obtain alternative clusterings via orthogonalization.

Yakmo implements robust, efficient k-means clustering with triangular inequality and smart initialization, while supporting alternative clustering outputs. The use of the triangular inequality allows k-means to skip unnecessary distance calculations, while the smart initialization by randomized seeding (k-means++) not only improves solution accuracy but also accelerates the algorithm convergence. In addition, you can obtain alternative clusterings via orthogonalization.

+ Show more

http://www.tkl.iis.u-tokyo.ac.jp/~ynaga/yakmo/