Changes between Version 3 and Version 4 of FAQ


Ignore:
Timestamp:
Feb 17, 2009 1:42:14 PM (13 years ago)
Author:
katyas
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • FAQ

    v3 v4  
    1 Q: Which SVM formulation does SVM-QP solve?
     1Q: Which SVM formulation does SVM-QP and SVMPath solve?
    22
    3 A: SVM-QP solves the 2-norm soft margin SVM classification problem, exactly the same formulation as is solved by
     3A: SVM-QP and SVMPath solve the 2-norm soft margin SVM classification problem, exactly the same formulation as is solved by
    44[http://svmlight.joachims.org/ SVMlight].
    55
    66-------------------
    77
    8 Q: How can I use SVM-QP?
     8Q: What is the difference between SVM-QP and SVMPath?
     9
     10A: SVM-QP is a Fortran implementation of a QP solver to solve one instance of a SVM problem. The implementation contains interior point
     11and active set algorithms.  SVMPath is a C++ implementation of the active set method in SVM-QP, which is also extended to produce the
     12entire regularization path of solution for a given range of regularization/penalty parameter values.
     13
     14-------------------
     15
     16Q: How can I use SVM-QP and SVMPath?
    917
    1018A: You can compile SVM-QP into a library and call it as a subroutine by passing it the data and the labels arranged into
    11 appropriate data structures. You also can set the kernel and other parameters for the problem. For details refer to the README file distribued with the source.
     19appropriate data structures. You also can set the kernel and other parameters for the problem. For details refer to the README
     20file distributed with the source.
     21SVMPath can be used as a callable library or as a stand alone code. It read the data from the input file in DOC format - same format as is used
     22by [http://svmlight.joachims.org/ SVMlight]. Various parameters for SVMPath are set in a parameter file.
    1223
    1324--------------------
    1425
    15 Q: How big are the problems that SVM_QP can handle?
     26Q: How big are the problems that SVM_QP/SVMPath can handle?
    1627
    17 A: This depends on the version that you use, the available memory and the size of the optimimal active set.
     28A: This depends on the version that you use, the available memory and the size of the optimal active set.
    1829 In Linux on and IBM (not-so-high-end) laptop we were able to solve the '''adult''' and '''web''' problems from the
    1930[http://www.ics.uci.edu/~mlearn/MLRepository.html UCI repository ] in a matter of minutes or even seconds. However,
    20 the number of optimal active support vectors (the examples that are exactly on the margin) did not exceeed 1500 in these tests. If the number of active support vectors is very large and is similar to the number of data points then SVM-QP
     31the number of optimal active support vectors (the examples that are exactly on the margin) did not exceed 1500 in these tests. If the number of active support
     32
     33vectors is very large and is similar to the number of data points then SVM-QP
    2134will probably be inefficient and will run into memory problems. However, we believe that such cases result in the
    2235overfitting of the data and, hence, it is questionable whether they should ever be solved.
     36SVMPath has very similar runtime to SVM-QP, but may be slightly slower due to C++. Also it may suffer from slowdown when there are too many breakpoints on
     37the regularization path.
    2338
    2439---------------------
    2540
    26 Q: I would like to try SVM-QP, but I don't want to spend too much time setting it up.
     41Q: I would like to try SVM-QP/SVMPath, but I don't want to spend too much time setting it up.
    2742
    28 A: Setting up SVM-QP may be easier that is appears from the first glance. Please contact the project manager for help.
    29 If you can  discuss your specific application we may help you estimate whether SVM-QP is the ideal solver for it.
     43A: Setting up  may be easier that is appears from the first glance. Please contact the project manager for help.
     44If you can  discuss your specific application we may help you estimate whether SVM-QP/SVMPath is the ideal solver for it.
     45In the future a Matlab interface is in the plan, please check back.
    3046
     47---------------------
     48
     49
     50
     51Q: What kind of problem does SINCO solve?
     52
     53A: SINCO (Sparse INverse COvariance selection) solves the same problem as is solved by [http://www.princeton.edu/~aspremon/CovSelCode.htm COVSEL]
     54and [http://www-stat.stanford.edu/~tibs/glasso/index.html Glasso]. In produces a sparse positive definite matrix which is an approximation of the
     55inverse of the  covariance matrix of a multivariate Gaussian model.
     56
     57---------------------
     58
     59
     60
     61Q: How can I use SINCO?
     62
     63A: SINCO has a Matlab interface (provided) which make it very easy to use in that setting. It can also be used as a callable C++ library.
     64
     65---------------------
     66
     67
     68
     69Q: How good is SINCO?
     70
     71A: SINCO is very much under development and testing at the moment. The overall performance is being evaluated. Please check with the project manager if
     72you want to use SINCO.