k-means classification

[asdkamps]

Hi,

I want to perform a k-means clustering with some 1D-arrays. The arrays are to be sorted into different groups which are generated during the clustering process. I already found that the NI Vision Development Module is something that could help me. Has anyone experience in k-means clustering under LV?
I'm very interested in your answers!

[PGriese]

Hy asdkamps,



The vision development module brings Classifiers with it. But it works on images and all of the classifiers there have to be trained. A k-mens could be done with customer classifiers, but I think it´s less work for 1D arrays to programm it without the vision development module.

I assume the 1D array is a vector in n-dimensional space and the distance functino D is euclidic distance?

Maybe the following paper is also of use, there k-means is used in LabVIEW.
http://edt.missouri.edu/Summer2005/Thesis/PogulaS-072105-T2944/research.pdf#search=%22%22k-means%22%20labview%22

[PGriese]

Hy,

If you give me an email adress can get you in contact with a developer who works on k-mean implementation, he will propably be able to help you.

[asdkamps]

Hello Peter,

thank you very much for your information. The link seems very promising but I will need some time to work it through.
Yes, you are right, the vector is n-dimensional (n=400-600) and the distance function should be euclidic.
Right now, my e-mail adress should be visible in the public user information. I'm very interested in discussing with the developer who works on k-means implementation.

Message Edited by asdkamps on 09-14-2006 02:11 AM