Implementing a parallelized kmeans algorithm with Cuda
Here are the tasks we need to go through
Step | Task | Status |
---|---|---|
1 | Write a program to generate data given the number of clusters, number of points and number of dimensions | Done! |
2 | Write the non-parallel 2D kmeans algorithm and verify it against OpenCV's code | Done! |
3 | Extend code to n dimensions | Done! |
4 | (practice) Get Cuda working | Done! |
5 | (practice) Calculate the squares of an array of number using Cuda | Done! |
6 | (practice) Do gaussian blurring with Cuda + OpenCV | |
7 | Figure out how to parallelize k-means with Cuda | |
8 | Write parallel code for k-means with cuda / verify against non-parallel algorithm | |
9 | Code cleanup and code review |
A bunch of useful links / resources found on the internet