In my point of view, to optimize the hyperparameters can not use standard
optimization techniques(or else it will become a parameters and cannot be
set empirically?) So some heuristic in brute force searching maybe a good
idea. I am thinking another heuristic to accelerate such process: maybe a
war
just an idea. what about a gridsearch using multidim optimization? As
in compute a heuristic and try to converge to an exact number
On Sun, Feb 10, 2013 at 10:03 PM, wrote:
> I have a pull request for randomized seaech but I need to update it as it
> is quite old...
>
>
>
> Ronnie Ghose
I have a pull request for randomized seaech but I need to update it as it is
quite old...
Ronnie Ghose schrieb:
>afaik yes. Please tell me if i'm wrong, more experienced scikitters :)
>
>
>On Sun, Feb 10, 2013 at 9:23 PM, Yaser Martinez
>wrote:
>
>> Any further development on this? Is a "brut
afaik yes. Please tell me if i'm wrong, more experienced scikitters :)
On Sun, Feb 10, 2013 at 9:23 PM, Yaser Martinez wrote:
> Any further development on this? Is a "brute force" grid search the only
> alternative to the problem of parameter selection for lets say SVMs?
>
>
>
>
> -
Any further development on this? Is a "brute force" grid search the only
alternative to the problem of parameter selection for lets say SVMs?
--
Free Next-Gen Firewall Hardware Offer
Buy your Sophos next-gen firewall be
2013/2/10 Olivier Grisel :
> I see that setup.py still does 2to3 translation in the build folder in
> that branch. Have you tried to get rid of it?
I didn't even try running it under Py3 yet. The problem is that
mock_urllib2 is replaced by something called UrlopenMock (for obvious
reasons), and I
Hi all,
I have trained a a Random Forest with 10K trees but I am not sure this is
the right number of trees for my datatset.
I'd like to tune the right number of trees using the oob score. However,
from the solution I see that the oob_score_ is just a float number. Which I
guess is the average of
2013/2/10 Olivier Grisel :
> 2013/2/10 Lars Buitinck :
>> Hi all,
>>
>> An update regarding Python 3.x compatibility.
>>
>> * I'm done rebasing Olivier's changes. I pushed the innocuous ones to
>> master, including some six-related stuff (and, accidentally, a change
>> that broke some tests, but th
2013/2/10 Lars Buitinck :
> Hi all,
>
> An update regarding Python 3.x compatibility.
>
> * I'm done rebasing Olivier's changes. I pushed the innocuous ones to
> master, including some six-related stuff (and, accidentally, a change
> that broke some tests, but that's been reverted). I managed to sa
Hi all,
An update regarding Python 3.x compatibility.
* I'm done rebasing Olivier's changes. I pushed the innocuous ones to
master, including some six-related stuff (and, accidentally, a change
that broke some tests, but that's been reverted). I managed to salvage
most of the changes, though not
a friend of mine organized a challenge to determine strategies for winning
in a game which simulates (at a very simplistic level) natural disasters.
he asked me to see if a collaborative solution could be faster and better
than the targeted solutions he is getting from people.
i've summarized the
+1. train_test_split is important
On Sun, Feb 10, 2013 at 12:21 PM, Gael Varoquaux <
[email protected]> wrote:
> Excellent idea. I would also like to include an important joblib fix. I'll
> try to do this tomorrow.
>
> - Original message -
> > Hey everybody.
> > I think it mi
Excellent idea. I would also like to include an important joblib fix. I'll try
to do this tomorrow.
- Original message -
> Hey everybody.
> I think it might be worth considering a bug-fix release.
> There was a problem with the nosetests and train_test_split on older
> nose versions (I t
Hey everybody.
I think it might be worth considering a bug-fix release.
There was a problem with the nosetests and train_test_split on older
nose versions (I think)
and a bug in MiniBatchKMeans for sparse matrices.
These just cherry picking these and doing a quick release wouldn't be
much trouble
14 matches
Mail list logo
