mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ted Dunning <>
Subject Re: Elkan triangle optimization for K-Means
Date Sat, 26 Mar 2011 20:35:51 GMT
Good idea.

On Sat, Mar 26, 2011 at 12:52 PM, Robin Anil <> wrote:

> I know that, what I was suggesting(by the term "decently modelled") was to
> write a benchmark which say keeps old distances and and allow elkan
> optimization to skip distance computation and see what is happening. He can
> time read/write and overall compute time using that class. He can also try
> it on different vector sizes. An experiment of that scale will a fairly good
> picture of what should be expected.

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message