On Wed, Oct 12, 2011 at 8:09 PM, Alexandre Passos
<[email protected]> wrote:

> Does anyone know how to implement parameter averaging without touching
> every feature at every iteration? With things like CRFs you easily
> have millions of features, only a few hundred active per example, so
> it's a pain to touch everything all the time. In the page he mentions
> that
>
>    Both the stochastic gradient weights and the averaged weights are
>    represented using a linear transformation that yields efficiency gains
>    for sparse training data.
>
> Does anyone know what format this is?

I'm not sure if it's the trick used by Bottou but this comment by
Peter should be helpful:

https://github.com/scikit-learn/scikit-learn/pull/162#issuecomment-1129889

Mathieu

------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2d-oct
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to