Implement a pre-processing phase + cluster initialization

polisquad / mpi-project

Implementation of the K-means algorithm with OpenMP and MPI

GNU General Public License v3.0

0 stars 1 forks source link

Open kjossul opened 5 years ago

kjossul commented 5 years ago

Ideas:

Remove noisy data
Sort data and divide it into L partitions: here we could use MPI to implement parallelism for each partition. This produces local clusters, therefore some sort of global reduction mechanism is needed for the final results.
Cluster initialization: here a good SO answer with popular initialization methods

kjossul commented 5 years ago