mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kim Falk Jørgensen <kim.falk.jorgen...@gmail.com>
Subject Re: Reg Algorithms.
Date Mon, 18 Mar 2013 14:33:12 GMT
Hi,

You could use the "one vs all" classification, where you basically
serialize the classifications asking for each category if its inside the
category or in some other. then you can use all the binary classifiers like
naive Bayes (http://en.wikipedia.org/wiki/Multiclass_classification). I am
not sure if that is what Danny Busch is also referring to.

Cheers
Kim



On Mon, Mar 18, 2013 at 2:56 PM, Danny Busch <danny@kurbel.net> wrote:

> Hi,
> typically, multi category classification can also be achieved by training
> multiple binary classificators. You loop through each of them and for each
> hitting classificator, you add a category.
>
> best
> Danny
>
> http://www.klickbrett.de
>
>
> VIGNESH S <vigneshklncit@gmail.com> hat am 18. März 2013 um 14:43
> geschrieben:
> > Hi,
> >
> > What are the best algorithms suitable for multi category
> > classification not binary classifcation.
> >
> > As Far as my understanding,Bayesian,SVD not suitable for multi
> > category classification..
> >
> > Please kindly help..
> >
> > --
> > Thanks and Regards
> > Vignesh Srinivasan
>



-- 
Best Regards

Kim Falk Jorgensen

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