On Thu, Sep 6, 2012 at 5:50 PM, Vlad Niculae <zephy...@gmail.com> wrote:
> I think that the "tweaks" our implementation has are vital for real world 
> use. However the perceptron is "textbook" and it would be nice to have a 
> simple way to reproduce the simple version. Is it just a question of init 
> parameters?

Probably that's what got me confused, I am used to thinking of
gradient descent algorithms in terms of perceptrons. But since you
already had SGDClassifier, it might have made sense to subclass it -
that ensures code reuse and also enough room to add fancy methods. Is
that the case?

Anyhow, in the SGDClassifier, where does the training happen? Is is
here - 
https://github.com/jaidevd/scikit-learn/blob/master/sklearn/linear_model/stochastic_gradient.py#L362

Also, what is meant by an unbalanced dataset?

>
> ------------------
> Vlad N.
> http://vene.ro
>
>
>
>
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
> _______________________________________________
> Scikit-learn-general mailing list
> Scikit-learn-general@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Scikit-learn-general mailing list
Scikit-learn-general@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to