Data process for large scale datasets

Kmeans: Overall, speed (single thread): Yael_kmeans > Litekmeans ~ Vl_kmeans

1.Vl_kemans (win10 + matlab + vs13 compile problem, but Win7 + matlab13 +vs12 can)

2.Litekmeans (direct use, single form faster)

Http://www.cad.zju.edu.cn/home/dengcai/Data/code/litekmeans.m

3.Yael_kmeans (multithreading) Compile time select Useopenmp=yes, matlab make file to add-fopenmp, otherwise cannot multithreading (will appear ignoring #pragma omp parallel). The NT value cannot be adjusted Yael_kmeans plus NT settings . For example:

Mex mex_sum_openmp.c cflags= "\ $CFLAGS-fopenmp" ldflags= "\ $LDFLAGS-fopenmp"

process:./configure.sh configuration, make, compile generic file, modify the make in Matlab, and then run the make file in matlab

https://gforge.inria.fr/frs/?group_id=2151&release_id=6405

OpenMP Programming: http://www.ibm.com/developerworks/cn/aix/library/au-aix-openmp-framework/

ANN:

1.Flann (compiled by tutorial)

http://www.cs.ubc.ca/research/flann/

Data process for large scale datasets