[med-svn] [libshark] branch upstream created (now 260213b)
Goswin von Brederlow
goswin-guest at moszumanska.debian.org
Fri Feb 6 11:55:47 UTC 2015
This is an automated email from the git hooks/post-receive script.
goswin-guest pushed a change to branch upstream
in repository libshark.
at 260213b removed old csv interface, killed warning in IHR?
This branch includes the following new commits:
new 4bcefd7 Moved actual library to trunk.
new fccf3de new solver structure for the Crammer&Singer SVM
new ceedcd9 new multi-class SVMs up and running!
new 35af980 Added alternative ctor that can consider only a subset of columns of the input data. re-committed b/c of new repo path.
new e12f6af git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@587 f06a09e3-3943-0410-8714-a4bb87ba143f
new 897a669 ASSERT macro fixed (puh!)
new df4d3af current libsvm versions also support multi-class files: small update to libsvm saver
new 78b5d44 also update libsvm saver from 0-based to 1-based class and feature indices
new 7828325 added loadLIBSVM, and updated SaveLIBSVM for consistency
new 1ed23cd Various Dataset constructors converted into Create... methods
new 2b97c04 Dataset syntax updated in all ReClaM tutorials
new 53a7128 minor adaptations to new constructor layout, and 2 small label conversion utilities
new 217a6af Fixed one of the ugliest bugs ever, of course, related to shrinking in multi-class SVM training. puh...
new 6e4b50f git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@599 f06a09e3-3943-0410-8714-a4bb87ba143f
new 99d1cc2 tiny adjustments, for overall consistency of dataset handling
new 9d8dfab git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@601 f06a09e3-3943-0410-8714-a4bb87ba143f
new df9ce32 fixed a bug in CMACMap
new 30e804f fixed the derivative of AffineLinearFunction
new eb3377f new dataset interface
new 14652ed bit fancier
new 3fabfeb for music paper
new 50dd2b6 fixed a tiny bug in libsvm-reader
new 56a0abb bug fixed in QuadraticProgram - finally!
new bb0b5e7 fixed again derivatives of linear Models. Didn't work for more than 1 output
new a4d0279 changed libname from Shark to shark
new 4df190b Reverted changes to CMakeFiles.
new 72a3a30 outputDimension = target.dim(1); needed for derived classes
new a1ade6f for ESANN paper
new d69827d new Dataset class, not tested
new 2f2140a [Shark trunk] Preparations and corrections for inclusion in debian.
new ee9cb48 FIXED: Compile-bug in CrossEntropy.
new 37a5b67 beta version of epoch-based crammer-singer solver
new 3807fa3 [Shark] REMOVED: XML parser with incompatible license. REMOVED: Debian block in copyright statement.
new 1b96801 [Shark] Fixed copyright issues.
new 36b626c Why are the QWT people change their API more frequently than Felix Magath the team lineup?
new a3280b5 new g++ smiles again
new bbb0b82 added an example for FisherLDA
new bdfa656 [Shark] FIXED: Bug in RMMEDA unit test. FIXED: Renamed method rank to sym_rank in order to avoid name clashes with std::tr1::rank.
new 431beaa git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@664 f06a09e3-3943-0410-8714-a4bb87ba143f
new 083f662 [QuadraticProgram] Removed reference to std::tr1::unordered_set.
new 5f99b73 SVMregression example repaired
new e071763 fixed derivative of Softmax - the derivative should not be input(j)*t but input(i)*t
new 28dfdf4 fixed ROC crashes.
new 615bf50 skeleton for multi class svm approximation
new 4d4ebf9 changes for multi-class svm approximation
new 2463e61 removed "false" parameter in resize after warning on Linux
new 9848f01 error function for McSvmApproximation
new 8d88b6c minor changes in FisherLDA
new 2030717 changes for McSvmApproximation
new abdefe7 McSvmApproximation
new cd630e6 also add a flush after each verbosity-2 output
new 4fac1d8 index shift
new 5c6786a numerical problems in QP solver fixed
new e5daf72 bugfix die Zweite
new 86a41a2 true positive rate added
new c195bab git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@774 f06a09e3-3943-0410-8714-a4bb87ba143f
new 2010fc2 - split all solvers into sub-files: QuadraticProgram.h now only includes all solver headers - added sparse kernel function (which owns its own sparse data) - added Utilities file, including SvmStatesCollection which can be used to i.e. later examine intermediate solver/svm states - further modifications of the ebcs machine
new 5ea4e1d forgot to remove now-oblivious QuadraticProgram.cpp
new 871fc05 added documentation
new a43933d git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@793 f06a09e3-3943-0410-8714-a4bb87ba143f
new 6277293 implemented ci's suggestion: remove stochastic component in cicl-target-noof-hits determination
new 181e1e8 balanced sampling for multiple classes
new b472197 balanced sampling for multiple classes
new d186fe0 - added some overdue ignores - made an attempt in windows cross-compatibility - fixed dynamic allocation for SvmStatesCollection
new d0c757e git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@829 f06a09e3-3943-0410-8714-a4bb87ba143f
new 78fdc9f conversion to one hot
new 284bd7c typos in doxygen
new 7304ab7 corrected typos
new 4697eac a second attempt in win-compatibility
new 64f9599 finalized the wall-clock timer cross-platform issue
new d7f7559 Preparations for bugfix release 2.3.3.
new 82c9cd4 [Shark2] Reworked the CMake setup.
new b59849c [Shark2] REMOVED: Old cmake directory.
new e7777db [Shark2] FIXED: Cleaned up the whole shark cmake setup.
new 54fe796 [Shark2] ADDED: #define NOMINMAX to prevent generation of min and max macros polluting the global namespace on windows.
new 56eb3bc small fix for documentation parser
new ce85315 [Utilities] FIXED: Compilation error for wall-clock timer under Mac OSX.
new 921d5b2 small fix for documentation parser
new 08312dd fixed my editor going whitespace-rampage
new bce9082 added ignores for example executables now generated in new place
new ea67d92 [Utilities] Fixed preprocessor macros
new 5763602 [Utilities.h] Mac OSX does not support cstdint, switched to stdint.h.
new 1450c40 [Shark2] FIXED: rt is not required on osx.
new 518c0eb [Shark2] UNIX AND NOT APPLE.
new 6e4fd6d [Shark 2.3.4] Adjusted patch version in CMakeLists.txt
new eaf600f 1 -> old tiny change for corrupt libsvm format as e.g. present in lasvm data 2 -> modified qt classification example to allow for true svg output (except the boundary and margin, ask for inkscape cookbook per mail if needed)
new c02b447 - incorporated jan's copy constructor and = operator - added some ignores
new 5073459 [Shark 2]FIXED: crowdedDistance now calls crowdingDistance internally.
new 2629727 messy code backup of mulit-class svm approximation (not working yet)
new 047c08f much cleaner code for multi-class svm approximation (still not working)
new 82cbb5b added a few ignores
new 548e3d9 small bugfix for McSvm operator=, found and requested by the RTCV group
new 0678834 Fixed an issue with the overmars-yap hanging in certain corner-cases.
new 0cd2d9e added new qt flag
new 72115fd fixed compilation issues with gcc 4.7.0 also fixed bugid 3554077
new e30815e moved trunk to tags and removed outdated shark2 version
new f90dd81 moved batch branch to trunk
new 305e111 up
new e5cbe4b renamed split again to splitAtElement
new f2cc115 tiny docs
new 184e192 up
new e20f882 output compiler
new bc3bd84 CI first reading
new 2580755 up
new 20e8ee6 up
new c024630 up
new ab8d6f3 up
new f4ab2b8 up
new 54a798a revised, CI
new b1eb758 up
new 7f3e8fd up
new 46346b7 up
new 32affa6 bug in linear SVM solver fixed, THANKS, OSWIN!!!
new 393b8ea removed unused typedef
new dd96f69 small updates
new 95291fa up
new 79f83cb up
new 8167bf4 small revision of tutorial text
new dddd39c updated data interface
new fd13f36 up
new 46d27c6 fixd dumb bug.
new 49777d4 restructured tutorials
new babee57 fixd intentation
new 6f134d2 tried to remove todos
new b5bf3ec git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@2612 f06a09e3-3943-0410-8714-a4bb87ba143f
new 927f03c - small typos - enable use of conf.py.in with python3
new 96cc825 RBM-Doku
new 8294c6f up
new 1216a63 error corrected and RBM-doku
new 823bfb3 RBM-doku
new c29a2d7 stray characters removed
new 806576c removed doxylink as contributed software, now leaving its installation as responsibility to the user. updated docs accordingly. cleaned up conf.py.in.
new d600574 hopfully fixd pca bug
new 7efc12f forgot lines in CMakeLists.txt
new 3576092 RBM-DOKU
new dc11204 https -> svn
new dea69b7 RBM-DOKU
new 9338ece fixed PCA and addd a new test case
new d8f4ee4 fixed bug in CART trainer, I asked Kasper to check it
new 623fe1d sucssfully updated the first part of the dataset tutorial
new ad890aa changed https to svn
new 7946840 doku
new 0b91018 up
new 42fb44c tiny revision
new 8a15bb9 reordered CMakeLists.txt a bit and splittet into tree parts
new 2f693c9 link corrected
new 477dd02 removed dataset constructor, added free function instead
new 7e2c358 Data distribution split into supervised and unsupervised cases, one class SVM example added.
new ee4923e fixed clustering bug
new fd804f7 forgot one file, sorry!
new 0c18c97 fixed a small bug in splitAtElement and got serialization to work
new 98b659c updated documentation
new a145d90 further cleanup of data docs
new 8e5d65e updated elm tutorial
new 3b0dac8 bug corrected, derivative threshhold was missing an entry for the regularization parameter
new ee6161d changed LibSVM Test a bitbefore new parser is integrated
new 4c19738 adjusted for new datset interface (hopefully correctly)
new dcd80ad updatd Libsvm parser and fixed a bug regarding dense vector proxies
new cb38cc1 fixed problem that sparse matrix rows are always copied when arguments for kernel::eval. huge speedup for sparse-SVMs
new 3c39d71 better parameter check
new a9091e0 comments only
new cea44ce Network component moved and replaced. This is still a stub, not nearly operational. Hang on :)
new 3503889 changed problematic output from examples to cout also other small fixes
new 8493c03 fixd bug regarding swap of elements
new 57a8429 just another stub
new a2206b7 web server component moved (again)
new 798501d small fix for clang 3.1
new 2aec901 updated shark packaging for cmake
new f8eaee0 fixed small bug in cmake file
new 668597d tested basic kernel matrices
new 35e4f91 fixed compile error on some compiler
new 01103b5 enabld multithreadd error functions...yay
new faca00f small improvements of documentation
new 36b9bdb typos corrected
new 13b0daf fixed compile error and a few warning on MSVC
new 2c3ce49 fixed small configuration bug in FisherLDA
new 1dcad3e fixed OpenMP on MSVC10
new dfef320 up
new a8d1852 incorporated better box constraint handling mechanism and cleand up DirectSearch a bit
new d5d7ab9 : -> ::
new efe9ce2 forgot a file
new fb45f6c added softMembership to LinearClassifier
new ad89815 comments updated
new 8c66990 new tutorial on how to write a kernel function
new 4e70f7f typos corrected
new 964ce6d made a function "const"
new 1312d5e up
new 8645ab8 applied small optimisation to the kernel matrix such that we now use the direct iterators to the elements instead of the dataset iterators. good improvement in performance as a result
new 3d069da fixed Bug in Matrix-Proxy and Benchmark test
new 319325b typos
new 35ae03e preliminary version of LASSO regression, still will compile errors, don't add to build system yet!
new 2501ce3 fixed compile error in Lasso
new b8b6e36 LASSO regression, now fully operational. And lightning fast :)
new 3bac87e allowed parsing class labels 1.0000 for LibSVM import
new 6ee05a2 corrected size() and element() for batch
new 76418e9 corrected size() and element() for batch
new 529cfcf added svn protocol
new 1626848 git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@2688 f06a09e3-3943-0410-8714-a4bb87ba143f
new 6f191ec RBM-Doki
new f48db84 RBM-Doku
new 2fda144 slight changes to LRUCache
new 2dfe584 wrote tutorial about writing objective functions, took the chance to chang the interface again. should be working. also update objective function tutorial
new ee9c8a1 rm errors/warnings under Ubuntu
new d2b6c66 moved slow to compile libsvm importer in library
new 4a56296 experimentation with failing unit test
new 27654b5 included LBFGS Algorithm
new f972c55 included LBFGS Algorithm
new 8b30c5f fixed broken build
new 548a27e corrected \file
new a327f8b updated bounds
new 45a7e3a editorial changes to the excellent tutorials
new 307ac9d editorial changes
new d7a6884 LOTS of changes, due to (*) new default implementations of interface functions, (*) changed INameable::name interface Testing done: everything compiles in debug and release and no (new) unit tests fail.
new 2aab85b up
new b711cf1 removed some warnings under Ubuntu / init order, size_t
new ab88270 test writes number of iterations
new 52024a1 up
new 2b85ad9 corrected docu rangedSubset(...)
new 38ff647 distance -> distanceSqr
new a29b90a merged the new binary svm solver
new 21fa86e changes in file headers only
new 33ca6d8 nothing serious :)
new 0a69865 tree nearest neighbors bug fixed
new e868714 must be tested, documented, etc.
new cb4ba1c was double
new dee8614 repaired
new eee97aa corrected wrong initial ifndef
new 23f0e2d fixed ugly bug in KMeans
new 8ec834f fixed warning in new Svm solver
new 7020eba tested CachedMatrix and found a small bug :(
new 5398e62 partially fixed performance issue in LinearModel with sparse arguments. it is now working for the McSvm-Case (dense output, sparse input)
new cff2968 kernel regression unit test improved and split into epsilon-SVM-trainer and regularization-network; added unit test for one-class-SVM
new 85f40b5 imcomplete xNES implementation removed
new 7c05f3a typo
new 3e9c9cf removed resize()
new 1d03a13 fix provided by Craig Brown
new 0f681af changes in comments only
new 5e573d9 changes mostly in comments only; criterium -> criterion
new 98528e5 imported ublas to shark and removed most of it's features
new 5233afa sorry :(
new e543954 altered the svm solver another time
new bcf1d7f quick fix for multi-class trainers; disabled shrinking for "max-loss" machines.
new 8598934 hacked in support for different regularization parameters
new 92c7265 changed order of regularization parameters...
new 4689516 reworked diagonal matrix
new 20132b3 changed shrinking another time
new 65d19c6 fixed an error and warning in debug mode
new 9fdfe05 fixed two more warnings
new f6048d1 fixed a LOT of warnings in ublas
new 74922f3 fixed last ublas warnings
new 0345529 removed another 3400 lines of ublas code
new 55e71f9 another intermediate step
new 860fca4 - improved docs - svnignore
new c26ccfe fixed small bug in Dataset
new 0b6e3eb deafault name added
new 1fd3605 fixed a small bug in the libsvm import and added a bit of parallelization
new 63efe17 added missing include statement
new b5a407a various cleanup
new c79d877 fixed compile error
new f867420 checked for and removed warnings on MSVC 10
new 6d72aa1 rewrote parts of the RBM package
new 61aff46 added .gitignore for git users
new 2f28d37 The netlib dependendy should now be obsolete.
new 05b12f9 added some preliminary documentation
new 32be625 generic "archive" wrapper for objective functions added
new 999c4d0 fixed and reactivated the TSP example program
new eaecebf First step towards including example code in tutorials. The tut2rst program does the macro replacement. The nearestNeighbor.tut contains the new tutorial source. KNNTutorial.cpp contains new marker comments.
new 6476a91 up
new 3010e91 typo fixed
new 18d7cf2 improved computation time for dense vectors
new ddc990d added new dense kernel optimization
new 69f22ac added new functionality to C-Svms
new 0e88399 small speed improvement of jaakkolla heuristic in th sparse hack. it's hacky though
new b516794 Added Sørens algorithmic improvements
new 21fe9e3 hopefully fixed an error in C++11
new 33b86f1 removed temporary files
new e5acff3 added AE tutorial, still needs review
new 2ec56ce up
new 9006460 small update of tutorial
new 1d17df3 size_t
new f802244 made output directory same as directory of the binary
new f41226e revision, e.g., added transformation functions
new acaf640 typos corrected, bug #70 fixed
new 2cb8dae typo in parameter description
new 560038d up
new ff8b97f add possibility to modify the linear part of an svm solution
new b71c193 sorry for compile error...cleaned up my mess. also fixed small error in nw example
new d783b0a fixed error in example
new 9b8ca98 fixed an error introduced by the change of the line search behavior
new d00025a fixed Loo error. sorry
new f75dffb added special case for rank deficieent SVM problems
new dd029a1 Fix Trainers_NBClassifier test failure
new d753c8c RBM Doku
new d8b921e fixed shuffle on VC08
new e708810 added squared hinge loss C-SVM and an alternative way to compute the bias
new 6deab6a fixed a small interface error in the distributions
new 8aaf28d fixed bug in hello world tutorial
new c425848 typo
new 3dc27f5 updated cross-validation
new e5c2f77 small fixes, mostly releated to documentation
new e6d965e added automatic code snippet inclusion to documentation
new 8b62185 some fuckups with git...
new e451a87 ported the first_steps tutorials to the new system
new 6d24621 ported tutorial code as well
new f639794 filename as option
new 5074d4a some fixes, started working on tutorials (little progress)
new 323dcee doxy
new 095785d fixed link to general optimization task tutorial
new 5652c41 bla
new bc152db fixed stupid bug in CMA
new d189fe7 fixed stupid bug in CMA
new e6ef56d updates of the tutorials, work in progress
new cdb4de0 updates, read files via command line, prepared for new tutorial system
new 4db2b27 ported some more tutorials
new 893562c typo
new 7a90be0 typos
new 1d4b107 typo
new 08be17c RBM doku
new 8a1bf94 up
new 76c6fb0 atlas headers
new 9b884fa mark_as_advanced
new bcb804b command added for including whole source files
new 3508b94 typo in comment
new ce1229f fixed out of source builds for doc
new ed146a8 started to rewrite linalg tutorial and fixed lda
new eedfe17 svnignore for new tut2rst, and some small corrections to tutorial and sphinx config
new 5b66fd0 added boost dirs to top level
new 7da97f2 commented message out
new 85cc5a5 web server tutorial stub added
new 847a545 various improvements to the linAlg tutorial
new dac1794 renamed tutorials .cpp->.tpp and automatically strip the tags from them with a conversion tpp->cpp#
new ac747d7 fixed a problem with broken ;
new 0744a98 fixd problems with regenerated examples after cmake run
new a5a9f5a updated .gitignore
new 3e9cc79 hopefully fixed the problem with shuffle
new 816f51d adapted filenames to point to the new examples
new 4fa5256 fixed tut2rst such that cmake now fails when the run was not successful
new 0e80031 updated elm tutorial
new 0f56eab updated ffnet tutorial
new deeecc6 fixed broken cma tut
new 6934e5e syntax error fixed
new 0cabda0 multi-task learning example added, still buggy
new 38ae85a some preliminary bug fixed, need to convert to MultiTaskKernel to batch interface
new 7e2c1ae ported and updated CART tutorial
new 946e5ce updated pca tutorial
new 4d27da4 fixd bugs in MultiTaskKernel
new d537b52 ported kmeans and rf tutorial
new f73dcb9 ported svm tutorial
new cad8b2a ported CSVMGridSearchTutorial
new 851e5ac ported Sparse Autoencoder tutorial
new 1975c25 just renamed the rbm module as it does not contain any code
new 71a8bd2 // we should move to Boost.Signals2 #define BOOST_SIGNALS_NO_DEPRECATION_WARNING 1
new 9c22549 small cosmetic changes
new cdf045c fixed case for svmProblems that for the subproblem of the workingset the second derivative was numerically 0
new b94c587 added new McSVM solver for box constrained svms (split from the big one)
new 5bb3373 added temporary fix for boxconstrained SVM problems
new 4f8c946 various fixes
new 435ee01 added /opt/local hint for ATLAS
new 40c0064 fixed various bugs in nearly unused shark code
new 0705227 changed naming scheme of variables in the trainer to match the standard m_ scheme
new f3e17f3 unified shrinking counter in QpSolver.h and fixed a hard to find bug on the way...
new adb583a solver now returns the unpermuted solution and uss the QpSolver interface
new 5aad7e8 random clean up and ported the LLW trainer to the new solver
new a845b25 cleanup
new fa021e9 update
new b81c5c6 * added more svn ignores * update of Shark.dox.in to current doxygen version * removed unneeded and misleading lines from conf.py.in * improved tutorials on doc system * tiny change to installation instructions
new 9ffe5e3 svn ignore
new 79ab811 * correct small bug in sphinx markup
new 9cfea62 * svn ignore for in-doc doc builds * included webserver tutorial in overview files * fixed bug in troubleshooting page
new 93a6979 * removed obsolete fullyweightedsum kernel from docs (was removed from code earlier)
new 1757238 clean delete of mkl-ckf.rst in preparation for tut-ification
new f5817e9 fixed interface error with Linar Mc Svm
new 26ea9ae eliminate some sphinx/doxylink warnings
new 0df3c3e fixed very small glitch when handling deactivated variables
new f73d88c fixed bug in radius margin quotient
new 3c595d2 filename changed
new 2713f6a removed probing
new 366f576 removed probing, added functions to get eiegndecomposition etc. to CMA
new 4480f22 minimal update of CMA tutorial, which needs revisionm; CMAProbe -> CMAPlot, has been revised
new f8043ae removed boost regex
new 54469d9 svn ignore for new mathjax version
new cd04ae2 fix bug in WeightedSumKernel.h (m_numParameters not initialized in ctor)
new 10c980b added some new cool operators to vectors
new c7ea4dc added new helper-function to partition a given number of points into batches
new 4db364d small cosmetic change to kernel expansions internals
new f240461 rewrote csv import
new 7f6b1e5 revised tutorial
new 31294a4 removed a warning in lassor egression - the timing code was unused
new a51920e oops forgot to fix the tutorials...
new 29e19aa removed bu just introduced
new 4dc8c69 added special case handling for gaussian kernels and sparse data
new 3f596ea docs: update boost link to persistent url
new 51a0351 fixed strange error. sorry
new a0b5d0f attempt to reduce mathjax timeout when offline.. not sure if working properly, but not working worse, at least
new 4a39e74 minor typos
new 354510b small doc improvements
new 51cd2af fixed small error in const ref handling
new 93754e1 used same label handling as for libsvm
new 2966127 added kernel kMeans
new ecd5659 repwrote LinearClassifier to be an argmax converter of a linear model. works great
new b55c998 fixed install of examples
new c009cb8 typo
new daea69c added stStructure interface to KernelExpansion
new ff01e84 removed default choice for wSS slection in CSVMTrainer
new 2289c90 cleaned up LinALg and removed unused algorithms. also fixed a bug in CSvmDerivative computation. Also moved everthing in LinALg to namespace blas
new 4e6be13 svn ignore
new 0f7363e updated mkl tutorial
new d8556fb renamed mkl tutorial
new 309c8f0 fixes small documentation things
new 21ade59 up dtlz2.setNoObjectives( 2 );
new 8bb1d58 warning in doc
new 0ecb504 added a stub on tut and tpp
new 1c5f077 fixed documentation warnings throughout sharks
new 7b3340b oops
new 59c161e RBM doku
new bcf49f0 convert csvm max-likelihood model selection tutorial to new format
new e2441c1 removed One-Hot encoding and fixed bug in BFGS
new 8b9b707 fixed a few smaller bug
new 3c6ffe9 undo accidental ci of rst file
new fe4a0a9 fix indentations, svn ignore
new da04c80 fixed compile error temporarily
new 90e7280 updated description and interface to regularization network trainer
new fdcb6ac removed thee automatic experiment interface for MOO experiments.
new b2eb1eb removed Logger
new 2091776 cleared Core from all now unused concepts
new 38a2b53 typo, but still does not work
new 1a8cd5a bug removed
new 086d2ef added a framework for LineSearch Algorithms. Reimplemented CG using a defined formula (with reference)
new 3ae1e2f brought MklKernel test up to date. still commented out in Test/CMakeLists.txt , though
new 035c02f reduced memory usage of RFTrainer
new 8003e47 added parallelization to RFTrainer
new dd528da git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@2922 f06a09e3-3943-0410-8714-a4bb87ba143f
new c7fa196 Input type of linear SVMs has become a template, allowing for arbitrary vector formats.
new 3a81699 added exampl code for brute force KNN CV
new b82af0f fixed small bug in the LinarSVM
new 0929615 Linear svms now use linear classifiers. also removed necessity to initialize model prior to training
new 0e5ca0c missing test
new 9af2645 web-server docs removed from core library
new aae13e9 small bug fix
new 6bf70ff sorry, forgot to check this in earlier
new 44aad42 added KernelClassifier and let all classifiers use it
new 2dc61b5 fixed bug in Dataset Iterator
new 78f4d50 headers adjusted, loss derivative for single element added
new 7475593 first version of kernel logistic regression added
new 2b1a0da training with offset enabled, number of epochs adjustable
new f55e8ee documentation improved
new ead52f8 bug fixes for KernelExpansion (m_outputs removed) and SGD (epochs initialized) changes in headers
new 669307b features are now allowed to start at zero
new 8e1ae41 better (?) constructors
new dabe642 KernelSGDTrainer not inherits IParameterizable
new 441b6e4 minor typos fixed in the docs
new f3d67e2 BUG FIX: libsvm import now throws if file opening fails. Ignoring this silently and returning an empty data set was actually quite nasty.
new e626459 comment header only
new 5a7b5e0 Archive wrapper for objective functions added. Most probably this will need revision.
new 3375564 Constraint handling bug fixed, thanks to Christian Aue!
new c42d215 added note on multi-threded libs
new 89bad10 improved KrnelSGDTrainer a bit
new 55d987d added new SimplexSolver for McSVMs which fixes a lot of the issues of the old solver. working
new 4fbaa63 aded a solver for systems without full rank, hopefully fixed LinearRegression by that
new 11fe246 fixed CSV bug and added tests for it
new 6bd545f added more tests, thus fixing ticket #59
new 2f30cbb span bound removed after debate on the mailing list
new 10877fc fixed compil error and test
new 7108f26 removed KalmanFilter
new 2a19dff cleaned up LinAlg further and moved lineseach algorithms to Algorithms/Impl. Linmin and Lnsrch removed as they are not needed
new 465c35d removed g_inverse and thus th last remaining invert* function. the several version of solve... are to be used instead
new c22f1b5 removed now unused ATLAS bindings
new 27008c7 inner iteration limit introduced, solves some (numerical?) problems
new 7ca66bf ooops, sorry, now it works
new 63ebc81 small interface change
new 58cbddc fixed compile error for export_csv
new 1784980 header cleanups
new 24469f3 made "official releae" advanved
new 27ab068 added fucntion to compute the complete kernel matrix at once
new 601e0ad added default batch sizee argument to libsvm import
new 419993f "five" changed into "six"
new 5e4f3eb test off by default
new 0571fc2 Fix the Dataset crash with clang on the Mac, and documentation improved.
new f4c838e up
new 8095e25 updated example buildfile such, that changed tpp files are automatically converted to cpp
new a92f540 no test fails
new 0e8cb37 no test fails
new 6d925f2 up
new 2419e75 typo
new 8bcf304 doc typos
new 3c0b573 attempt 1 at solving these f**king cmake issues
new 7c6b966 fixed tpp2cpp target dependencies
new d6c7064 small changes regarding typs and casts
new 8329fce svn ignore, tiny variable naming update
new cf7d043 changed attribute table
new eda10cb GLPK error handling fixed
new 78d411f tried to fixa few warnings and updated KernelHelpers.h
new 31bfd81 up
new e72ba00 made csv import easier to use - when a whitespace is given as a separator this counts as no separator, so there is no need for a 0 as argument anymore. Also fixed a bit of typos...
new 0a556f1 fix (or evade via ignore) remaining doxygen warnings
new 9ca61a4 version 0 of lasso toturial added
new 8bbaa56 git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@2989 f06a09e3-3943-0410-8714-a4bb87ba143f
new 19279ac git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@2990 f06a09e3-3943-0410-8714-a4bb87ba143f
new db0c086 Versatility of learning tutorial fixed, all code moved to example programs.
new 181088a Worked over statistics tutorial and example program.
new a5a9a7d git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@2993 f06a09e3-3943-0410-8714-a4bb87ba143f
new 776544b improved LibSVM import to use less memory
new 063a858 added measurement of wallclocktime to timer, which is now the default
new 9823e9e svn ignore
new 09db532 - let Csv test also test comment functionality. - tiny doc clarification
new da4ee43 Oswin's noalias fix ported from the LinAlg branch to the trunk for immediate benefits
new 729be02 bad typo
new 7195de8 up
new 04cd2cc typo
new 15a77c6 rewrote LinAlg completely
new 2a4dfc4 fixed comment handling in csv.h
new dcee6f3 git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3017 f06a09e3-3943-0410-8714-a4bb87ba143f
new 7f69ee4 simple "make uninstall" added
new cabd927 git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3019 f06a09e3-3943-0410-8714-a4bb87ba143f
new 3c690b4 namespace of isnan fixed
new 12f9b67 corrected typo
new c045ced added operator *= to dense vectors and matrices allowing for elementwise multiplication. replaced element_prod in shark by * or *=
new c1d9551 fixed compile bug for older boost versions
new 0e1ced3 optimized Gaussian RbfKernel derivatives
new 6b108c1 more missing includes
new 819845f more missing includes
new 18c512b added appending of datasets
new c5be9e5 fixed an issue regarding serialization of Classifiers
new 0a8bc41 - svnignore - added repository version to doxygen's 'The documentation for this class was generated from' ... - small typos in datasets.rst
new 1f5c021 timer problem fixed
new 871ec71 two new tutorials added
new fcd2455 examples for tutorials modified/added
new 90f8616 header
new 45fdd09 bugs fixed
new a7b28a5 more fixes
new 62027d1 fix of ugly "& &" --> "&&" bug
new 7af4a30 buggy (?!) third parameter removed in csv_import for labels
new b640802 small cosmtic changes
new 19c1ae5 added second derivates for losses
new dca5e4c added proper prior to LDA
new 2278c75 added a bit of numerical accuracy to scaling of C
new 34eb585 another timer fix
new 8fe523b git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3043 f06a09e3-3943-0410-8714-a4bb87ba143f
new f791dad added an ATLAS binding for inner_prod
new e041c68 improved handling of nested proxies
new e87a81c fixed small bug in sparse matrix transformations. deactivatd failing tests and declared as won't fix. Fiing that would either be stupid or slow - even in cases where it wouldn't require a change
new c42c445 fixed two errors on ubuntu: a missing include and a numercial problem with PCA with verz small eigenvalues - they are now ignored as they are only numerical noise
new 1cbd69d fixed small bug when adapting C while using cross validation(no current result is invalidated). Also removed a specialisation that is not necessary
new dd285ca fixed a bug in jaakkola heuristic which occured when two identical points had different labels. Also rewrote it so be a lot faster when used with ATLAS. still needs some better optimization
new e908378 HDF5 fix. CMake will now test if the existing HDF5 lib has the correct executables installed.
new 0de4c88 fixed serious bug in LibSVM WSS. Does not effect final results, and does not effect gaussian kernels. Only runtime is affected. Wrong matrix diagonal was chosen.
new b8532d4 made CrossEntropy numerically more stable. Also changed the starting step with of KernelSGDTrainer to be more reasonable
new 6dc93ce up
new 38f4b89 up
new b9c6596 fixed dysfunctional nearest neighbors test.
new 6496cd5 further fix which fixes the learning rate for high values of C
new f0769ec clang warning fixed
new 3505606 Add missing flag for Clang 5.0 on OSX
new e283ba0 small changes in documentation removed information about how to contribute / code review system
new d3bd495 git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3061 f06a09e3-3943-0410-8714-a4bb87ba143f
new 7763070 macro renamed
new c5e87a2 added function for numerically stable calculation of soft_max(x) = log(sum(exp(x)), where x is some vector
new ce1e95d changed JaakkolaHeuristic API such that original and faster algorithm is default
new db2f118 added AIS and other algorithms to estimate the partition function
new 44c5fb9 * fixed tooltip bogus content with new doxygen versions (>1.8.5) * svnignore * some typos * small doc improvements
new a323402 changed order of include files
new 5228550 up
new e685f03 added initial random number tutorial
new 12ea436 up
new 81e33d0 isnan -> boost
new 8e42e17 changes constructor such that test work again
new 5a42830 fixed probelm when uni / discrete is called with min=max; changed file encoding
new ea9c05f added comment, removed useless argument from distribution constructor
new fe06dc1 MultiVariateNormalDistribution API updated
new 702147a made CD and estimation of partition function multithreaded
new 098ce1e fixed small bug introduced by too small type
new ab87804 restructured code a bit to allow estimation of several algorithms from the same sample
new c1990dc fixed small problem that too many threads are created when there are not enough batches
new 3a52888 added burnin time
new 051461a added the full propper AIS as proposed by NEAL
new 7733c63 added multi threading again
new eb8732b fixed broken example
new 9a27a64 changed softPlus/soft_plus to use floating point precision. had to tune tests to the new numerics
new f85d275 fixed temeprature handling and renamed algorithms
new 4dedaa4 GNU linear programming toolkit removed as a preparation for license change
new b940465 changed license
new 6a72af2 up
new f697d6c fixed legal thingy that prevented CPack from building
new 8448883 removed GLPK.cpp as it was forgotten..
new ea26d24 added mean classifier
new eec013b rewrote RFClassifier to use MeanModel
new 24e7032 update header template
new 8c66770 update header template
new 9300847 add tool for replacing headers. will execute it this evening.
new 4f44b49 Update existing headers to LGPL.
new 7efef07 update of header replace script
new c8e0262 added news about LGPL
new 909ad00 up
new b35b52a up
new a2f6076 up
new bc73e30 Remove all file tags
new 81d2a66 adapted for OPenSuSE
new 6be08d3 add comment on shared libs
new 4d23c8a hopefully fixed vector for bool arguments
new 6b28883 added a stub for a weighted error function
new c18f9c6 partially added code from Kang
new 1c0a620 fixed a bug with reading in files with single integral values
new c417371 updated
new 325d753 up, typos
new 9de5da2 Visual bug
new b29b8cd added attribute counter
new 57da1ed added feature selection to dataset
new f404b7d minor typo
new eef662e typos
new 0fd6a3a bit of clean up
new a9e593d various clean up and added flip-thestat sampling
new c72ac7c added support for accessing columns / features in data sets
new 2df9370 various fixes and simplifications throughout the RBM library
new 7995be5 - svnignore - tiny typos and corrections in docs, sent from jacob
new 86887da added HingeLoss Loss-function
new e9efdf5 added squared hinge loss
new 1892ea9 cleaned up MarkovChain Tests
new f372038 implemented vector valued min max to single scalar values as in max(vector,0.0)
new aab45f0 sorry for compile errors
new f41cd24 - added better capabilities to recreate splits from indices: added index vector type to class signatures - tiny: added second level of verbosity option to grid search
new e425cf1 tried to fix normalized kernel test (was unused before). removed now unneeded old kernel tests, they are all in new files by now
new 24380b5 reactivated two older tests and got them to work again
new 7aed15f removed FitnessTraits
new ddaf485 fixed small bug in LDA
new 4895dd5 added psilonHingeLoss for Regression
new 0b78d28 added a bit of doc to the losses
new 110418f partly rewrote DirectSarch
new cace467 sorry for the compile errors in the example code
new 35af291 added missing include file
new 07dba12 add libsvm-like wrappers
new bb6c63f remove small typo
new 22b878b remove wrappers.
new 49895c2 git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3139 f06a09e3-3943-0410-8714-a4bb87ba143f
new a25dd0b git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3140 f06a09e3-3943-0410-8714-a4bb87ba143f
new e29d59a "inline" added to remove clang warning
new 53b3d2a 1/2 added - gives correct results but is somehow odd...
new c7574e2 up
new 8445f6d got DirectSearch back to compiling and cleaned up a lot of things
new 783c96a updated CMA a little
new 1c98ac7 I broke PT and fixed it again. Also added a way to store the energy differences for running chains using a wrapper for PT
new da07193 rewrote ElitistCMA
new 24d176c forgot to update file header
new 3d1a5e6 removed two unused files
new eadf401 oops
new b21cd41 removed more unused files
new 24a58fa added iterator constructor to vectors
new e93a4a2 further clean-up of direct search. moved a few deprecated components in Save before they get deleted. Merged all elitist cma parts into the CMAIndividual and rewrote MOCMA algorithms to use this
new 2fb48cf missed a file
new fee3fbc removed empty dirs
new 546f9f1 up
new ef0413d further rewrote DirectSearch
new 0c01e20 Fixed warning about signed-unsigned int comparison
new 6c051c3 git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3159 f06a09e3-3943-0410-8714-a4bb87ba143f
new 77f0801 unified MOCMA and ElitistCMA code
new 930f92a made MultiVariateNormalDistribution simpler
new 5028848 further cleaned up MultiVariateNormal and added MultiVariateNormalDistributionCholesky which uses a cholesky decomposition
new fbee7fb minimal test of default variant
new a93d2e4 first checkin of reworked data tutorials
new 71758ce markup fixed
new 822474c git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3166 f06a09e3-3943-0410-8714-a4bb87ba143f
new 06f30c8 more on data tutorials, this time with a bug :(
new 0113485 fixed dataset tutorial
new 643387e git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3169 f06a09e3-3943-0410-8714-a4bb87ba143f
new c56a8cf data import/export routines renamed, examples + tests + documentation updated
new 25286f5 git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3171 f06a09e3-3943-0410-8714-a4bb87ba143f
new 6d213be forgot one file
new af949ea git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3173 f06a09e3-3943-0410-8714-a4bb87ba143f
new 2b210c1 added cholsky update
new 9522374 added author/date
new a10c912 added parallel mini batches to the RBM package
new d5c570f removed AbstractFactory as well as unit tests for it
new 385374a removed plotting function
new fc4dc6c removed Derivative.h
new cd7ef79 removed Statistics.h
new 3cabecb removed docs to removed classes
new a743e4b updated ignored files in the doc
new 8b222fd removed SharkDefs - not needed anymore
new 3065c81 moved a test file tool to the Test folder
new 1705780 cleaned doxygen documentation a bit by removing template specialisations and deciding that detail/* and Impl/* should not be documented. Also moved stuff around to use this
new 6f13c4f redo of commit that removed Statistics.h
new 12fddd3 removed the Batch bindigs for boost::fusion::vector, they are superseeded by the other methods and thus useless
new dee7c3c removed SearchSpace and IsVector to make things simpler
new da9a151 fixed a broken test and cleaned up a small thing which allowed me to remove CopyConst. Also figured that SharedVector is some remains of some old thing we don't ned anymore -> can die
new b74b31b moved Fuzzy to addons
new d1f3f5c added missing documentation of statistics
new a1de6b9 oops
new b7ba561 error in gitignore lead to ignoring the examples
new e56e92c - svn ignore - added quickref directory, it's overview and the first page - added corresponding links to it, updated html templates
new 36225b9 Oswin's comments incorporated
new 1b1400b git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3196 f06a09e3-3943-0410-8714-a4bb87ba143f
new ba966ec (too stupid to proof read...)
new c4e31a8 added the first part of the SVM approximation
new d680283 documentation slightly altered
new f59b749 merged changes
new 1a2efe8 git-svn-id: https://svn.code.sf.net/p/shark-project/code/trunk/Shark@3201 f06a09e3-3943-0410-8714-a4bb87ba143f
new f4170c6 Refactor matrix classes
new 37b5602 fix for the includes. part 2/42
new d14bce7 added test for the box constraint handler
new eee42e6 small improvements to KernelBasisDistance - added a method to return the computed beta value
new 6e7c6b6 cosmetic fix
new 1a30b08 added a Test for PenalizingEvaluator and fixed a bug
new bccd53b hotfix.. part 3/42
new b04b261 added Dominance relation Tests and small fixes
new 7833174 fixed two reported bugs, one crippling performance, the other crippling the MOCMA. added unit tests and fixed two other smaller bugs. MOCMA has now a unit test which passes
new fac408c checked ZDT implementations and benchmarked the MOCMA with two more functions
new 5159760 add a special kind of cache that just makes sense for kernelsgd
new 6cd3d0c checked and cleand up benchmarks again and added even more tests - now also for SteadyStateMOCMA
new fc6d446 checking the other functions as well and cleaned up the code
new 22870cb added unit test for RealCodedNSGAII and fixed small compilation bugs. ZDT2 fails however
new 49ffeee whoopsie
new 957cabd small cosmetic fixes and made tests a bit faster
new 8057499 fixed a bug in automatic return type promotion when the argument was a scalar in binary operations. now vector<int>*double->vector<double>
new 3e10c5d added example for experimental usage of MOCMA
new 2d08b7d added an example for an MOO Experiment
new f540e13 unified mu_plus_lambda and mu_komma_lambda selection, it now fits in the general interface
new 469cc48 clean up, improved a test and removed a dysfunctional one, which only pretended to be a test
new 860a476 add vector include to avoid compile error on old mint 14
new 2aecf28 replaced RBFNet by RBFLayer for better reusability
new 004b775 added input derivative for linear model. this is needed to create a RBFNet using RBFLayeer and LinearModel
new 3899612 added possibility to only train one part of the concatenated model
new c855144 added a model representing a convex combination based on the lienar model. the main differences are the constraints used to enforce the convexity property
new 75dd815 added first draft of negative log likeelihood function. This is not stable as we explicitely compute first p(x) and then log(p(x)). as p(x)->0, this becomes unstable
new fc69123 removed bug in computing the angle of the ellipses
new c7eeccc typo
new 237b936 insert function to set sigma
new 499df56 changed to unix format
new 03982af 1st attempt to fixe scaling
new 7050ce6 added counter
new b7315e8 deactivated for now the fast covariance update as it is numerically unstable
new 345d5f7 added numeric safeguard to covariance matrix update, checked that it now runs finee with 200 dimensions
new 5f71e57 unused variable
new 470552e updated CMA-ES to match the 2011 tutorial version
new 9e40f68 up
new 2b7f43f added efficient rank 1 update for cholesky factors
new 6756385 added new, fast, intelligent cholesky update which does not require the inverse
new 0264467 a test was failing after the ellipsoid function got changed
new 0983f77 changed the Evaluator interface a bit so that it is possible to evaluate groups of individuals. This might allow us later to introduce noise handling
new da11656 we need boost 1.48 at least
new 8a25105 fixed small fuckup during another fix... thanks francois-michel
new b3db74e minor change (mainly for testing, so ignore)
new e49f700 add boolVector/Matrix..
new 121d610 "bug fixed" - or rather unit test adapted :/
new b744700 changed separator form string to char
new 7d44091 small fixes for documentation
new 6bf8a4d ref to RBFN
new 2765f37 fixd problems with dynamic linking on linux
new 23c9f4c added an evaluation example for the CMA
new f0b5ada made an example on how to train a mixture of gaussian unsupervised using KMeans and MaxLogLikelihood. Works on the faithful dataset, so i am sure it will always work...
new 26bbcb2 oops
new 21a51e8 Fixed some bugs and added some small feature for title lines in Csv.h
new f387c3c oops git failed me
new c0786e4 added missing kernel matrix includes
new 7c3a386 removed bad implementation of Noisy autoencoder error
new c1b5fb7 made it possible to set the initial step size
new a4b285c added distribution which samples image patches from images
new 3748212 added a testing function whcih checks that the weightedDerivatives method produces similar results as the individual calls
new 582f589 optimization for WeightedDerivatives not to call derivatives neadlessly
new 8b3d177 added helper function to interpret vectors as matrices more easily
new 7e262cb added model which corrupts its input by gaussian noise. usefull for denoising autoencoders
new 4d81863 improved temporary proxy handling
new 9d06325 rewrote FFNet and fixed a bug that lead to a crash when the bias was not used
new bb5ee68 make ACF and shrinking optional
new 083a021 allow non-const access to basis of kernelexpansion
new c6667f2 rename PrecomputedMatrix, minor typos
new 4598986 1. fixed small bug in that the maxIteration count was notused in one version of CG. 2. replaced Gram-Schmidt orthonormalisation for creation of random rotation matrices with househodler reflection algorithm, which is numerically more stable - this is now good enough for benchmarking numerical routines in high dimensions - it is also faster.
new 7b6b98e added a unit test for eigensymm which is at least a little bit more sophisticated. Also made a bit of the test code slightly faster and changed an unobvious noterror to an obvious correct piece of code
new c92e532 1. changed dynamic/static once more. now it should actually be working. 2. added bindings to ATLAS-LAPACK if ATLAS is build with full lapack (new build option for now, until i find a good way to query this)
new 5fc0caa moved the rank one update into the MultiVariateNormalDistributionCholesky
new 42f01e4 added new weighted datasets
new 2aca420 added the possibility to query the batch sizes for repartition
new f7bbd23 small fixes to the Weighted Dataset
new cfadc64 rewrote problem constructor to support for individual weights and C values for each class
new 3abb1c1 added interface for new weighted trainers
new a3c7434 added support for weighted CSVM training
new a398b35 added bootstrapping
new e0da5fe added cod for the multinomial distribution
new 96675f4 added another test
new 08994df small fixes and added querying functions as for normal datasets
new 20c2173 added weighting for LDA. Only problem is, that i can't get the unbiased estimates right
new bbafbf3 uncommented a test case..oops
new bec62ec removed DataObjectiveFunction
new d2a8e1e fixed small possible compile error
new 03ddac1 updated documentation a bit
new 02ad80e added function which automatically rotates another (probably axis aligned) function
new 73080c8 up
new 4a37eb0 added triangular matrix-vector product: triangle_prod(A,x) which computes x=Ax
new 2df3fef fixed a bug that Aydins build server found. I promise, i will never change a line in that ugly code
new 14f0454 added faster implementation for trsv in certain cases
new 4a6bca5 fixed HDF compilation warning
new 6f6c5db added loss for sequences and removed the derivative interface for AbstractLoss
new 631bfdc add first version of kernelBudgetedSGDTrainer
new 171a871 fix errors and docs
new 1dedda1 add projection maintenenace strategy, missing any tests
new fd7d3a5 update doc for projectbudget..
new d98b5f2 revert unintended checkin
new 3771fcb add label order class
new e498174 fixed initial step size of CMA in high dimensions
new e121df4 added two Large scale CMA variants. need documentation/tests but working
new be12b02 make valgrind and vs2013 happier than before
new fa2abfc fixed learning rate
new 09dfa68 definitions of cigar and discus where switched
new f4b04ff reenabled hand-modification of eigenvalues
new 71a0450 fixed bug in active covariance update
new 2f02f93 changed definition of beta such that now beta=0 is feasible for gaussian layers. accordingly it is now possible to hav a canonical definition of AIS etc which use equal temperature spacing. These methods are added for easier usage
new b929753 some changes to add momentum and make some hyper parameters more reasonable
new 28e95ab up
new b6a776a up
new 3735c7e small updates
new 2deb6da improved documentation
new 8a43397 typo documentation
new afe3560 added broader support for regularizer
new 682fb44 added the huber loss
new 481337e added Tukey's Biweight loss for robust regression
new b3d3a4c added tutorial to train a deep neural network on a simple toy task. Added various helper functions to make this easier
new 896b28b added tutorial for deep ffnets and changed a few other tutorials by removing lying parts/updating them
new 45059bd fixed
new 69ac84a small doc fix
new 3df09a8 minor bugfix/update
new 1d270d1 randomly swap loci
new 480e2d8 added min function...
new 8c578cb added regularisation methods to NoisyErrorFunction
new 0291faa added TiedAutoencoder (autoencoder with tied weights)
new b2d441d fixed LinAlg headers
new 24dfb10 added ZCA normalisation
new d6ed07a added Dropout
new c07e276 added 'Impulse' noise for denoising autoencoders on binary data
new 82b3cfb added also simple autoencoder with the same interface as TiedAutoencoder (and sharing much code unfortunately...). Tsted and works
new 96c4132 added Rprop with full weight backtracking
new dbc1b9d fixed svd problem
new f4f296e fixed small bug in tutorial when reading the datast. Thanks at Michiel
new cbeba1b remove NegativeClassificationLogLiklihood and updated tutorials to reflect this. Had to add a quick fix for SigmoidFit
new 0a0d0f4 mac rpath policy
new f875690 test for version
new 4a3b2ea make code more c++
new c6498b5 brought the Sparse-AE tutorial closer to the current state-of-the-art implementation
new de38344 removed the choice of ouput neurons for autoencoders
new be9a9a1 -undid the bad decision to have no output neuron type in Autoencoders -brought the SparseAEtutorial and the SparseFFNetError even more in line with current development by replacing FFNet->(Tied)Autoencoder -updated the documentation and updated our list of classes and methods
new 979111a added basic autoencoder tutorial
new 72b1b6f added denoising autoencoder tutorial
new e814dce added RBM tutorial and fixed some compilation bugs. Sorry for the late commit, my local git repo was broken...
new 076833c rm warning because of double (())
new ad8aec4 added documentation
new 3649d2a updated doc
new 0a53789 capitalization
new 236bf82 up
new 860df66 up
new a7df977 ignore
new 1cc4af8 small revisions
new a2bd702 test, added loval js conf file
new c47c598 up
new b036eb1 - expanded on chris recent changes to change the latex \vec command to print bold symbols instead of arrowed ones. In detail, add local mathjax config file, load it, and also load it for fallback local code. Plus all the same for doxygen pages, too.
new 67241b1 - repaired previous hacks for vector redefinition in mathjax for strictly no absolute paths (e.g. allowing copy to diku server with docs built elsewhere). seems to work
new 7c59623 updated tutorial with probability of samples section. also improved the softmax model as to also work with only a single input
new c5970ed added mask to TwoNormRegularizer
new 9f33e38 add componentwise sin/cos
new 1dc08f2 added helper functions
new 5866950 added fix in case that the highest index is available and a separate index is givn
new bde098c added matrix sin/cos for completeness
new 774b5d5 various changes
new 2f8236a typo in documentation
new 31875d7 fixed missing function problem
new c46f4bc bug corrected
new a4daeb3 fixed reported compile errors on VC2010
new 80d3b80 fixed the VD-CMA Algorithm. Worst problem was an error in the paper. I think it is now working fine, at least the convergence looks reasonable
new f544128 fixed bug in tutorials page
new 8ac10d2 whoopsie
new f04dfe0 fixed openmp issus with VC2013 as in ticket #80
new 81da003 fixed numerical error with NegativeAUC
new 26d3697 added new batchwise cross validation folds
new b9e6c87 fixed small bug in tutorial
new a3574ab fixed compilation problems with boost 1.57 beta(which does still not compile because of serialization)
new 3169558 removed unusd components
new bcc650d newlines
new 21a8fa8 Added the trigamma Function. Wow!
new db73a88 Added a trainer for the so-called reinforced multi-class SVM.
new f7826ea unit test and example added for reinforced SVM
new e08c1bd redid approximation of trigamma and added tetragamma function.
new 0fbf6dd typo
new dff461d 1/distance weights added to nearest neighbor models
new eda44fe change test suite names
new bd9744b change test suite names, rest
new 4aa7671 fixed noalias(x)*=2 failure and added a matrix_set on the way for Data<RealMatrix>
new 7f3d3f8 added ModelKernel class which computes a kernel with a model to map the inputs into another space
new e6bbd47 missing includes added
new 5930593 changed name to reflect name of inner function
new e7c509e added optimisation for column_major matrices when an outer_prod is applied
new 8bdaea3 various optimisations(and a small fix) for ElitistCMA
new d4b6e4b after i made another test more strict it turned out that this class didn'tset its proper flags...
new 5b3f78c fixed possible error with parallel Dropout
new dc7b7b0 added new triangular matrix class with packed n*(n+1) storage, also cleaned up on the way
new 00df7d8 fixed warnings and activated a missing test
new dcec81d added push_back for vectors
new 7ee37ed add coverage/memcheck
new 9fcfe3d update test cmake files
new 4fd4ab5 hide coverage, if mac was broken, this should fix mac build
new 5e2fadc fix stupid merge error. someone should ban me from using svn. or we move to git.
new a9e1b87 try to fix the stupid bug. its late, i need vacations
new cbee577 linear reinforced multi-class SVM (RM-SVM) added
new 24aa144 added scalar add/subtract for vectors
new 34c4ff0 missed a header file
new c7d22e6 added regression capabilities to KTA
new a7db251 added multi thread support for KTA(making the class even uglier)
new ac3bdb9 additional constructor the soft nearest neighbor classifier
new 37ddd71 added atanh
new ff6c03d oops
new 96e5512 remove bogus and unnecessary file
new e477559 small -> tiny
new 719f998 fixed a number of warnings in release mode
new 309d3b0 fixed another warning
new 8237562 added functor for fitness sorting
new 7ac27d6 small fixes in docu and added operator+= and -= for elementwise addition of a scalar. + and - where already there
new fb8bef2 update copyright date
new 3f86291 added new benchmark functions
new 387518b added alpha parameter to Cigar
new f33a5c8 fixed bug in CMA-ES. sigma was updated to early. Thanks Tobias.
new dcdb9ab add valgrind results folder
new 801de94 minor fixes in gridsearch tut
new 260213b removed old csv interface, killed warning in IHR?
The 918 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails. The revisions
listed as "adds" were already present in the repository and have only
been added to this reference.
--
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/debian-med/libshark.git
More information about the debian-med-commit
mailing list