none
parallel k-mean cluster RRS feed

  • Question

  •  

    Hi all,

    I try to parallel the k-mean cluster algorithm to get a faster one with mpi....

    i work on real data set, the data contain 1030 row (instances) and 8 column

    First time - master send data and center to other processors and the all processors make distance computation between all row and three centers to decide the class of each row. 

    then the processors will be send the class of each row to master

    the now compute the new centers and the send there to other all processors fo re-compute the distance.......etc

    but i have problem .... when the compute the new centers i send  the new centers to all processors but i feel that all processors does not receive

    the new centers?? so what is the problem

    can you help me ??

    my code here>>

    http://www.4shared.com/file/mFqFrEQB/MPI_p-k-mean.html

    thanks


    • Moved by Dana Groff Thursday, June 2, 2011 9:28 PM Moving to the correct forum (From:Parallel Computing in C++ and Native Code)
    Tuesday, May 17, 2011 6:52 PM