Changes between Version 12 and Version 13 of WikiStart
 Timestamp:
 Feb 12, 2009 2:33:01 PM (13 years ago)
Legend:
 Unmodified
 Added
 Removed
 Modified

WikiStart
v12 v13 1 = Welcome to the SVMQPhome page =1 = Welcome to the OPTIML home page = 2 2 3 3 == Introduction == 4 5 OPTIML stands for Optimization methods in Machine Learning. Right now this page contains three subprojects: 4 6 5 7 SVMQP ('''S'''upport '''V'''ector '''M'''achines '''Q'''uadratic '''P'''rogramming solver) is a software package that solves 2norm soft margin support vector machine [http://www.kernelmachines.org/ classification problem]. The problem is formulated as a convex QP of the following form … … 13 15 where {{{ e }}} is the vector of ones, {{{a}}} is a vector of lables (1 or 1) of the data, and {{{C}}} is the panalty parameter associated with the violation of the margin constraint. 14 16 15 SVMQP is designed for largescale SVM problems. The underlying algorithm is an active set method for convext QPs. [http://www.research.ibm.com/people/k/katya/incas.pdf/ Here ] is the paper describing the algorithm and containing computational comparison with [http://svmlight.joachims.org/ SVMlight]. The software also includes an interior point17 1. SVMQP is designed for largescale SVM problems. The underlying algorithm is an active set method for convext QPs. [http://www.research.ibm.com/people/k/katya/incas.pdf/ Here ] is the paper describing the algorithm and containing computational comparison with [http://svmlight.joachims.org/ SVMlight]. The software also includes an interior point 16 18 SVM solver which is designed for problems where Kernel matrix is (approximately) low rank. The program constructs the low 17 19 rank approximation and solves the approximate problem by the interior point method. The approximate solution can then be passed 18 20 to the active set solver as the warm start. The underlying algorithms are described [http://www.ai.mit.edu/projects/jmlr/papers/volume2/fine01a/fine01a.pdf here]. 19 21 20 Th e currently available versions is a Fortran 77 code whichis designed to trade memory for22 This versions is in Fortran 77 code and is designed to trade memory for 21 23 efficiency. Currently, this is the most time efficient version of SVMQP. 22 A memory saving versions, which is somewhat slower will be available soon.23 A beta versions can be obtained by contacting[http://www.research.ibm.com/people/k/katya/ Katya Scheinberg].24 A memory saving versions, which is somewhat slower is available in the C++ version (see SVMPath). 25 In case of questions, contact [http://www.research.ibm.com/people/k/katya/ Katya Scheinberg]. 24 26 The current version is desinged to ba called as a subroutine where the burden 25 27 of parsing the data and arranging it into appropriate data structures lies with the user. 26 28 27 The C++ version of the software is under development by [http://web.mit.edu/belloni/www/ Alexandre Belloni]28 and will be available in the near future. Additionally the C++ version will include the ability compute a path of optimal solutions for any given range of parameter {{{C}}}.29 29 30 2. SVMPath is the C++ extension of SVMQP. Additionally to solving the SVM problem for a given value of parameter C, the C++ version includes the ability compute a path of optimal 31 solutions for any given range of parameter {{{C}}}. This version is strictly in the development stage right now. The manual and other supporting materials will be coming soon. 32 Please check back or contact the authors. 33 34 3. SINCO ('''S'''parse '''IN'''verse '''CO'''variance) is the C++ software with a Matlab interface which solve the sparse inverse covariance selection problem 35 {{{ 36 max K ln det{C} tr(AC) \lambdaS.*C_1 37 C in S^{p x p} 38 }}} 39 where {{{K, \lambda}}} are positive scalars, {{{A}}} is a {{{p x p}}} symmetric matrix, {{{S}}} is a {{{p x p}}} nonnegative matrix and {{{C}}} is the unknown {{{p x p}}} 40 symmetric positive definite matrix. This software is written in C++ and has a Matlab interface which is provided. A brief description of how to use the code is available via 41 comments. A more detailed manual and the algorithm description is forthcoming. 30 42  31 43 32 44 == Background == 33 45 34 SVMQP is released as open source code under the [http://www.opensource.org/licenses/cpl.php Common Public License (CPL)]. It is available from the [http://www.coinor.org/ COINOR initiative]. [http://www.research.ibm.com/people/k/katya/ Katya Scheinberg] is the COIN project leader for SVMQP.46 OPTIML is released as open source code under the [http://www.opensource.org/licenses/cpl.php Common Public License (CPL)]. It is available from the [http://www.coinor.org/ COINOR initiative]. [http://www.research.ibm.com/people/k/katya/ Katya Scheinberg] is the COIN project leader for OPTIML. 35 47 36 You can obtain the SVMQPcode either via subversion or in form of nightly generated tarballs. To get the tarball, go to the COIN [http://www.coinor.org/download/source/SVMQP tarball directory], and look for a file like svmqp3.x.x.tar.gz.48 You can obtain any of the softwares available in OPTIML code either via subversion or in form of nightly generated tarballs. To get the tarball, go to the COIN [http://www.coinor.org/download/source/SVMQP tarball directory], and look for a file like svmqp3.x.x.tar.gz. 37 49 38 Individual files can also be obtained from the svn web interface (see the "Browse Source" button above). The SVMQP distribution can be used to generate a library that can be linked to one's own C++, C, or Fortran code. SVMQP39 can be used on Linux/UNIX platforms and Windows. 50 Individual files can also be obtained from the svn web interface (see the "Browse Source" button above). The SVMQP and SVMPath distributions can be used to generate a library that can be linked to one's own C++, C, or Fortran code. SVMQP and SVMPath 51 can be used on Linux/UNIX platforms and Windows. SINCO can be used from C++ and from Matlab via the available mex interface. 40 52 41 As open source software, the source code for SVMQPis provided without charge. You are free to use it, also for commercial purposes. You are also free to modify the source code (with the restriction that you need to make your changes public if you decide to distribute your version in any way, e.g. as an executable); for details see the CPL license. And we are certainly very keen on feedback from users, including contributions!53 As open source software, the source code for OPTIML projects is provided without charge. You are free to use it, also for commercial purposes. You are also free to modify the source code (with the restriction that you need to make your changes public if you decide to distribute your version in any way, e.g. as an executable); for details see the CPL license. And we are certainly very keen on feedback from users, including contributions! 42 54 43 55 In order to compile SVMQP, certain third party code is required (such as some linear algebra routines). Those are available under different conditions/licenses. 44 56 45 We provide th is programin the hope that it may be useful to others, and we would very much like to hear about your experience with it. If you found it helpful and are using it within our software, we encourage you to add your feedback to these wikibased webpages, [wiki:SuccessStories Success Stories].57 We provide these programs in the hope that it may be useful to others, and we would very much like to hear about your experience with it. If you found it helpful and are using it within our software, we encourage you to add your feedback to these wikibased webpages, [wiki:SuccessStories Success Stories]. 46 58  47 59