Conversation
|
In this case using linear interpolation for images looks better, and you can also add an interactive covariance matrix, see if the scatter is more informative in 3D |
|
scatter looks pretty much the same in 3d I think interactive cov is fine to add for the gallery demo but is prob too much for medium article to explain |
I think this looks better! Easier to see the numbers
I think it shows off the cool features of fpl though, it shows that covariance is a good measure in this case. |
The covariance matrix is not that informative in this case, we already have an example with covariance I think for this demo it is okay to just keep it to kmeans (show people something different) mnist_cov.mp4 |
|
@kushalkolar should be gtg |



Simple kmeans clustering example using MNIST dataset
kmeans.mp4