Nice work. This should make big improvements possible.
On Tue, Jul 28, 2009 at 7:16 AM, nfantone <[email protected]> wrote: > After all those modifications, kMeans now runs noticeably faster. > Running the whole thing locally -in a pseudo-distributed cluster-, > every iteration is taking me about ~7 minutes to complete using the > very same dataset I uploaded and posted some days ago, which used to > last 3 hours. Again, this is not meant to be a final, closed solution > at all. But, I think it could very well serve as a first step towards > that. > -- Ted Dunning, CTO DeepDyve
