I updated the gist with the example in __main__ to create the plots as in
http://www2.isye.gatech.edu/statistics/papers/05-25.pdf, figure 1.
Bon weekend
J
2012/4/11 Jaques Grobler
> haha okay, I'll give it a swing!
>
> J
>
>
> 2012/4/11 Alexandre Gramfort
>
>> > The simulations that are done i
haha okay, I'll give it a swing!
J
2012/4/11 Alexandre Gramfort
> > The simulations that are done in the last mentioned paper by Ming Yuan
> and
> > Yi Lin,
> > http://www2.isye.gatech.edu/statistics/papers/05-25.pdf, they find
> that the
> > NG seems to do
> > generally better than the LASSO
> The simulations that are done in the last mentioned paper by Ming Yuan and
> Yi Lin,
> http://www2.isye.gatech.edu/statistics/papers/05-25.pdf, they find that the
> NG seems to do
> generally better than the LASSO (figure 1)
if you can reproduce this figure using my gist I pay you a beer :)
I
@Olivier - no problem :)
@Alex
Appart from trying it ourselves to see how it fares, here're some findings:
The simulations that are done in the last mentioned paper by Ming Yuan and
Yi Lin,
http://www2.isye.gatech.edu/statistics/papers/05-25.pdf, they find that
the NG seems to do
generally bette
Le 11 avril 2012 11:53, Alexandre Gramfort
a écrit :
>
> it seems to me that it might be a good addition to the scikit if can
> convince ourselves with examples that it does better than a Lasso.
I agree.
Thanks for the write-up Jacques.
--
Olivier
http://twitter.com/ogrisel - http://github.com
> The algorithm proposed in this paper, is rather similar to that of the Lars
> LASSO, but with a complicating
> factor being a non-negative constraint on the shrinkage factor. (See eq. (2)
> in this paper)
> Once you've computed your shrinkage factor, you basically have your
> regression coeffic
Here's a wee summary on the non-negative garrote (NG) i pieced together:
The original non-negative garrote from Breiman (1995) is basically a scaled
version of the least square estimate.
Basically take a OLS estimator and then shrink that estimator to obtain a
more sparse representation.
The shrin
> Does it give it extra consistency properties? e.g. unbiased estimates?
could be … Jaques will explain this to us tomorrow :)
He's watching the talk on video-lectures :)
Alex
--
Better than sec? Nothing is better than
To continue with funny algorithm names, can we have the top moumoute online
natural gradient algorithm in scikit-learn :) ?
http://nicolas.le-roux.name/publications/LeRoux08_tonga.pdf
Mathieu
--
Better than sec? Nothing i
On Tue, Apr 10, 2012 at 02:52:04PM +0200, Alexandre Gramfort wrote:
> it has a rescaling step like Adaptive Lasso but using OLS. The
> positivity is just to impose the same sign of in coef_ and coef_ as
> obtained with OLS
Does it give it extra consistency properties? e.g. unbiased estimates?
G
> Yes, basically the non-negative garrote is a non-negative Lasso, if I
> understand it correctly. Thus if your priors are that your model is
> sparse, and with only positive weights, the non-negative garrote is the
> right estimator.
no :)
it has a rescaling step like Adaptive Lasso but using OL
On Tue, Apr 10, 2012 at 02:44:56PM +0200, Jaques Grobler wrote:
>This paper mentions
>We also show that the nonnegative garrote has the nice property that
>with probability tending to one, the solution path contains an
>estimate that correctly identi es the set of important variable
> What the benefit of non-negative Garotte?
unclear to me for now. I'm still reading on the topic. But if somebody
can pitch in I'm interested.
Alex
--
Better than sec? Nothing is better than sec when it comes to
monitor
hahaha @Olivier 's Garotte cake
This paper mentions
We also show that the nonnegative garrote has the nice
property that with probability tending to one, the solution path contains an
estimate that correctly identi es the set of important variables and is
consis-
tent for the coecients of the im
> > What the benefit of non-negative Garotte?
> To cook a non-negative Garotte cake?
Definitely tastier than a negative Garotte cake.
--
Better than sec? Nothing is better than sec when it comes to
monitoring Big Data ap
Le 10 avril 2012 14:39, Gael Varoquaux a écrit :
> What the benefit of non-negative Garotte?
To cook a non-negative Garotte cake?
--
Olivier
http://twitter.com/ogrisel - http://github.com/ogrisel
--
Better than sec? No
What the benefit of non-negative Garotte?
G
--
Better than sec? Nothing is better than sec when it comes to
monitoring Big Data applications. Try Boundary one-second
resolution app monitoring today. Free.
http://p.sf.net
hi,
as soon as we have Immanuel's branch with positive lasso [1] merged we
could have a non-negative
Garotte in the scikit. A quick gist (hopefully not too buggy):
https://gist.github.com/2351057
Feed back welcome and if someone is willing to cleanly merge this …
Alex
PS: I've added the snippe
18 matches
Mail list logo
