Hi everyone,
There's a workshop in machine learning open source software at NIPS this
year (http://mloss.org/workshop/nips13/).
I think we (the scikit-learn team) should submit something. I know
scikit-learn has been presented a couple of years ago at this same
workshop, but it is IMO a major mach
On Thu, Aug 22, 2013 at 09:32:13AM +0200, Nelle Varoquaux wrote:
>workshop, but it is IMO a major machine learning toolkit, and we should
>not hesitate to resubmit something.
One of the questions is: who will be there? I might, or might not be
at NIPS this year (depending on whether our pa
Hi all,
I have been trying to search a equivalent python package for Bayes Net
Toolbox https://code.google.com/p/bnt/
Of which I can specify a hierarchical hidden markov model (hhmm) and
train it with data. Is there any python package available for this
kind of task?
--
王硕
邮箱：shuo.x.w...@gmail.
Hi Björn,
On Wed, 2013-08-21 at 22:20 +0200, Björn Esser wrote:
>
> my name is Björn Esser. I am a packager-sponsor and provenpackager from
> Fedora and want to rpm-ify scikit for Fedora and RHEL. My first contact
> to this project has been 'NelleV' on IRC, who recommended to write to
> this ml
Like Olivier said, libsvm and liblinear are heavily patched and
scikit-learn wouldn't work with the upstream versions.
If bundling them is unacceptable, I guess maybe packaging our forks
individually as libsvm-sklearn or something similar would be a solution,
but I think it would be confusing. Al
2013/8/22 Vlad Niculae :
> Like Olivier said, libsvm and liblinear are heavily patched and scikit-learn
> wouldn't work with the upstream versions.
>
> If bundling them is unacceptable, I guess maybe packaging our forks
> individually as libsvm-sklearn or something similar would be a solution, but
Am Donnerstag, den 22.08.2013, 10:41 +0200 schrieb Tadej Janež:
> Hi Björn,
>
> On Wed, 2013-08-21 at 22:20 +0200, Björn Esser wrote:
> >
> > my name is Björn Esser. I am a packager-sponsor and provenpackager from
> > Fedora and want to rpm-ify scikit for Fedora and RHEL. My first contact
> > t
Am Donnerstag, den 22.08.2013, 12:17 +0300 schrieb Vlad Niculae:
> Like Olivier said, libsvm and liblinear are heavily patched and
> scikit-learn wouldn't work with the upstream versions.
> If bundling them is unacceptable, I guess maybe packaging our forks
> individually as libsvm-sklearn or some
2013/8/22 Björn Esser :
> Am Donnerstag, den 22.08.2013, 12:17 +0300 schrieb Vlad Niculae:
>> Like Olivier said, libsvm and liblinear are heavily patched and
>> scikit-learn wouldn't work with the upstream versions.
>
>> If bundling them is unacceptable, I guess maybe packaging our forks
>> individ
Am Donnerstag, den 22.08.2013, 13:46 +0200 schrieb Olivier Grisel:
> 2013/8/22 Björn Esser :
> > Am Donnerstag, den 22.08.2013, 12:17 +0300 schrieb Vlad Niculae:
> >> Like Olivier said, libsvm and liblinear are heavily patched and
> >> scikit-learn wouldn't work with the upstream versions.
> >
> >>
I'm sure you will hate this suggestion, but what about creating a text
file/command line "interface" to existing machine learning executables.
advantages:
a) no problem with data copy: the executable loads data from file (you
don't need to keep in sklearn)
b) most ML algos are available from comma
2013/8/22 Björn Esser :
> Am Donnerstag, den 22.08.2013, 13:46 +0200 schrieb Olivier Grisel:
>> 2013/8/22 Björn Esser :
>> > Am Donnerstag, den 22.08.2013, 12:17 +0300 schrieb Vlad Niculae:
>> >> Like Olivier said, libsvm and liblinear are heavily patched and
>> >> scikit-learn wouldn't work with t
2013/8/22 Sean Violante :
> I'm sure you will hate this suggestion, but what about creating a text
> file/command line "interface" to existing machine learning executables.
> advantages:
> a) no problem with data copy: the executable loads data from file (you don't
> need to keep in sklearn)
> b)
I agree with Olivier's remarks.
lightning supports a rudimentary command-line interface [*] but that's
because I want to make it easy to non-Python users to try my algorithm on
their data.
Mathieu
[*] http://www.mblondel.org/code/mlj2013/
On Thu, Aug 22, 2013 at 9:10 PM, Olivier Grisel wrote:
2013/8/22 Olivier Grisel :
> 2013/8/22 Björn Esser :
>> Can you please tell me a bit more about what / why these were modded?
>> If we cannot unbundle them, I need to have some further infos to get
>> some exception granted for them. :)
>
> For the libsvm binding the main reason it to get both the
BTW, the advantages of scikit-learn's approach over text-file based
programs are also briefly discussed in our recent paper:
http://staff.science.uva.nl/~buitinck/papers/scikit-learn-api.pdf
Mathieu
On Thu, Aug 22, 2013 at 9:20 PM, Mathieu Blondel wrote:
> I agree with Olivier's remarks.
>
> l
Am Donnerstag, den 22.08.2013, 14:22 +0200 schrieb Lars Buitinck:
> 2013/8/22 Olivier Grisel :
> > 2013/8/22 Björn Esser :
> >> Can you please tell me a bit more about what / why these were modded?
> >> If we cannot unbundle them, I need to have some further infos to get
> >> some exception granted
On 22 August 2013 14:38, Björn Esser wrote:
> Am Donnerstag, den 22.08.2013, 14:22 +0200 schrieb Lars Buitinck:
> > 2013/8/22 Olivier Grisel :
> > > 2013/8/22 Björn Esser :
> > >> Can you please tell me a bit more about what / why these were modded?
> > >> If we cannot unbundle them, I need to ha
2013/8/22 Sean Violante :
> a) no problem with data copy: the executable loads data from file (you don't
> need to keep in sklearn)
Quite the contrary. What if only raw data (text files, JSON, etc.) is
on disk, and you still need to do feature extraction on it? Then you
need a pipeline of a featur
Nelle,
That's a great idea! I hadn't been planning on attending, but because NIPS
is so close to me this year I think I may be able to make it if there were
a compelling reason to be there. Is anybody else from the core team
planning to be there? Any thoughts on what sort of talk/presentation to
Hi,
It is more than likely that I will be there this year - given the
reviews of our paper, I would be surprised if it was rejected.
What sort of talk would you have in mind Nelle?
Gilles
On 22 August 2013 16:13, Jacob Vanderplas wrote:
> Nelle,
> That's a great idea! I hadn't been planning o
sorry for coming late to the party.
yes -- I have not "externalized" anything (liblinear/libsvm) besides
joblib. The branch which tracks main releases of scikit-learn
is https://github.com/yarikoptic/scikit-learn/tree/releases
and from it https://github.com/yarikoptic/scikit-learn/tree/dfsg
provi
Hi Shuo Wang.
There seem to be quite a lot of packages, if you google "python bayes net".
I have never used any of those, though.
Depending on what you are after, you might want to look into pymc:
http://pymc-devs.github.io/pymc/#
or my pystruct:
http://pystruct.github.io/
PyMC does mcmc inference
23 matches
Mail list logo
