Closed GoogleCodeExporter closed 9 years ago
Original comment by pierre.raybaut
on 25 Jan 2009 at 8:38
I've been reading the svm.py module from libSVM and I found out that because of
some
issue they had with their Python binding they do not exclude zeros from their
sparse
data representation, therefore, the sparse representation is only there to make
the
whole thing totally inneficient.
In their website, specifically http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/ ,
there is link for a dense data representation version of libsvm which
approximately
40% faster, however, without a Python binding.
If you guys are considering to include libSVM in the Python(x,y) distribution,
a
binding for the dense data libSVM version would be ideal.
Cheers,
Albino
obs.: Thank you guys for this wonderful project!
Original comment by albinoad...@gmail.com
on 16 Feb 2009 at 11:04
I would also like to support this request for a machine learning library,
preferably
libsvm which is the most popular out there.
Concerning the the version for dense data, although it would be great to have it
added, the sparse version can still serve in many cases without significant
problems.
Considering also the fact that it is currently available in python, I believe
that it
can be directly and easily added to the core distribution, leaving the dense
version
for a later stage.
Thanks
Original comment by chpsal...@gmail.com
on 27 Apr 2009 at 10:33
The new OpenCV version (2.0) Python binding fully supports the Machine Learning
library
that ships with OpenCV (includes AdaBoost,Support Vector Machines, Random
Trees, Neural
Networks, KMeans clustering, among others). Please, consider adding the OpenCV
2.0
python binding in the next update of Python(x,y).
Albino
Original comment by albinoad...@gmail.com
on 16 Dec 2009 at 7:14
OpenCV 2.0 has been added to next release:
http://pythonxy.googlecode.com/files/opencv-2.0.wr1.1.0_py26.exe
Original comment by pierre.raybaut
on 30 Mar 2010 at 9:44
Thanks a lot! :)
Original comment by albinoad...@gmail.com
on 12 Apr 2010 at 3:08
Original issue reported on code.google.com by
albinoad...@gmail.com
on 24 Jan 2009 at 11:44