Monday, October 03, 2005 5:30 AM
I've been playing around with K-Means clustering lately and found an example in C#
here. I worked on some optimizations to this code and made the clustering work on multiple threads which gave it a 40% - 60% decrease in execution time depending on how many vectors are being clustered, how many clusters are generated, and how many dimentions there are per vector.
Here is the source for the optimized K-Means algorithm.