SHOGUN  4.1.0
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Modules Pages
FAQ.mainpage
Go to the documentation of this file.
1 /*!
2 \page faq Frequently Asked Questions
3 
4 \li Q: I am puzzled, shogun interfaces to so many languages but which language
5 and shogun interface should I use? A: That depends a lot on your taste. I
6 personally consider the modular interfaces (python_modular, octave_modular) to
7 be "best". Here best means very flexible and easily extensible. However, in case
8 you just want to train a single SVM with a single or multiple kernels all of the
9 static interfaces are sufficient. And well, of course you should be using \b
10 python http://www.python.org :-)
11 
12 \li Q: I've found a bug, where should I report it? A: Either report it in our
13 bug tracker https://github.com/shogun-toolbox/shogun/issues or ask on the
14 mailinglist.
15 
16 \li Q:Do I need CPLEX to use multiple kernel learning?
17 A: No, as of version 0.7.0 you won't need cplex if you want to learn the weights
18 in front of the kernels. However, to enable Multiple %Kernel Learning \b with
19 CPLEX(tm) just make sure cplex can be found in the PATH. For standard 1-norm
20 multiple kernel learning (MKL) the GNU Linear Programming Kit (GLPK) version at
21 least 4.29 or CPLEX is required. For general p-norm MKL with p>1 it will work
22 nonetheless.
23 
24 \li Q: Is it multiple kernel learning when I use many kernels? A: No, a plain
25 combination of features/kernels will remain a plain concatenation. Only in case
26 you learn the kernel weights you really do MKL. In the static interfaces you can
27 issue \verbatim [W]=sg('get_subkernel_weights') \endverbatim and check whether
28 all weights W are still 1.0.
29 
30 \li Q:Does shogun compile under windows?
31 A: Yes! With cygwin 1.7.1 the cmdline, python, python_modular and octave
32 compile cleanly. However, octave_modular won't work since the swig version in
33 cygwin is outdated - if you want that interface you need to compile swig
34 manually. In addition neither R and matlab interfaces do currently compile:
35 cygwin does not yet include up-to-date mingw packages. As none one of us use
36 windows, and things (in external dependencies) break frequently, feel free to
37 submit patches to cygwin. Ohh and one hint: use
38 \verbatim
39 make DESTDIR= install
40 \endverbatim
41 to install (etc.). A plain ./configure;make;make install will fail since
42 repeated slashes (e.g., '//usr/local') in cygwin are problematic but in posix
43 systems simply ignored.
44 
45 \li Q:How does shogun do its memory management? A: As does python, we use
46 reference counting internally, i.e. objects holding a reference to another
47 object should increase the reference count of the object they are referencing
48 and decrease the counter when they finished using the object. It should be noted
49 that loops (e.g., object A holding a reference to object B and vice versa) are
50 not detecting and may thus create memory leaks. However, this scenario can so
51 far be easily avoided - just don't create a combined kernel that contains itself
52 as a subkernel ;-)
53 */

SHOGUN Machine Learning Toolbox - Documentation