Actually, from my point-of-view, the current interest & buzz surrounding
Mahout is divided in two broad categories:
"scientists", who are interested in the theoritical setup of machine
learning algorithms based on map-reduce technologies
"engineers", who have a technical or functionnal need that can be solved by
machine learning tools such as Mahout.

I include myself in the 2nd category, but I tend to think that to
successfully answer a business need I have in mind, I have to get a better
grasp on the theoritical concepts underlying in Mahout. This mailing-list,
and the book, as well as other ressources (such as Programming Collective
Intelligence), are all great ressources.

(sorry for this off-topic & non technical rant on this technical ML...)

2010/5/20 Sean Owen <[email protected]>

> It looks like a big classifier. I'm guessing it's going to be pretty
> great. It's such a natural fit with their infrastructure.
>
> Building out these underlying map-reduce implementations is great, but
> in the end the useful thing is to wrap that up in an API that is
> oriented towards a particular application.
>
> That's the hard part too -- how do you make such a general process
> applicable to many types of data in many contexts with one general
> framework or API? It looks like they may provide many particular
> domain-specific APIs on top of this engine. That's what would be
> really great.
>
> ... and I don't doubt people, on this list perhaps, are also
> interested in building a series of specializations on top of Mahout to
> orient it to specific domains and use cases. That's an interesting
> business!
>
> On Thu, May 20, 2010 at 8:43 PM, Florent Empis <[email protected]>
> wrote:
> > Hi,
> >
> > This is a new service by Google of supervised learning
> >
> http://googlecode.blogspot.com/2010/05/bigquery-and-prediction-api-get-more.html
> >
> > Anyone has thoughts on this? I'd say it seems pretty limited as of now,
> but
> > the details are scarce...
> >
> > Sorry for this slight off-topic... :)
> >
> > BR,
> >
> > Florent Empis
> >
>

Reply via email to