[opengm] annotated tag debian/2.3.6+20160814-1 created (now be14bec)
Ghislain Vaillant
ghisvail-guest at moszumanska.debian.org
Thu Sep 1 09:19:42 UTC 2016
This is an automated email from the git hooks/post-receive script.
ghisvail-guest pushed a change to annotated tag debian/2.3.6+20160814-1
in repository opengm.
at be14bec (tag)
tagging 4e9c1d97d92f9a48aa7debb52f9792cfee3890b5 (commit)
replaces debian/2.3.6+20160131-2
tagged by Ghislain Antony Vaillant
on Wed Aug 31 09:34:01 2016 +0100
- Log -----------------------------------------------------------------
opengm Debian release 2.3.6+20160814-1
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=4hk2
-----END PGP SIGNATURE-----
Carsten Haubold (33):
Fix gurobi library dependency include for tests
Rename opengm::Parameters (aka ModelParameters) for learning to opengm::learning::Weights, and graphicalmodel/parameters.hxx to graphicalmodel/weights.hxx
Fix header include guard for weights
fix learnablefunction renaming, method was called "paramaterGradient"...
Add generalized hamming loss with small test
Add cplex backend for quadratic solver. Compile learning test if cplex or gurobi is available.
Merge commit '4c4667669a21f362e693c66fd2bfe38a7f303771'
Introducing loss parameters
Fix call operator of sum of experts which did not use the mapping of local to global weight indices yet.
Remove unnecessary write to labelOffset_ in LUnary constructor
Make sumOfExperts available from Python
Remove Loss from StructMaxMargin template parameters. Update parameter object, and actually pass it to the optimizer. Use infParameter in Oracle. Use Gurobi OR Cplex as INF in test.
bring StructMaxMargin learner to python. needed inlining of all bundle optimizer methods.
use #ifdef instead of #if in test_learning.cxx
Add visitor that defines the learn methods for each Learner and all Inference methods.
Update real world example, add Cplex to available inference methods for solvers.
Link opengm python learning module against CPLEX and/or Gurobi
Fix tests after learnablefunction.hxx and l_potts.hxx were removed (but probably still present in installation destination of some developers)
Update dataset constructor on python side to set up the weights vector as well
Store best weight vector in dataset when done.
Adapt python wrapper to changes in struct max margin learner
Update structure of maxLikelihood. Export to python, run from real example 2
Embarassingly simple example: one-variable model, with max likelihood learning
Only build SelfFusion if QPBO is enabled
Find package OpenMP and build the subgradient ssvm test with those flags set
Allow to build SubrgradientSSVM without OpenMP, and link the learning python module against openmp
Merge branch 'master' of bitbucket.org:jkappes/opengm-learning
some learning python tests
Add RebindGM to LPCplex2 so that it can be used for structured learning
Fix an #elif statement to find the cplex backend for the struct-max-margin solver
Add LpGurobi2::RebindGm to allow learning with Gurobi solver
Use IndexType instead of UInt16 in LearnableUnary, to allow large numbers of features
Also install header files with extension .h -- needed for structured learning solvers
Constantin Pape (1):
Intersection Based and 3-cycles
DerThorsten (122):
learnable functions in python seems to work
Merge branch 'master' of github.com:joergkappes/opengm-learning
learnable functions in python seems to work
setting initial values from numpy array works
added lunary to python
Merge branch 'master' of github.com:joergkappes/opengm-learning
new learning module
mychanges
merged
added const to param ref in grid search learner
inital version of grid search learner in python
BAZINGA bitches, its working
Merge branch 'master' of github.com:joergkappes/opengm-learning
added py helper function
improved python learning
improved python learning
fudwefrew
Merge branch 'master' of github.com:joergkappes/opengm-learning
python
python loss param object
Merge branch 'master' of github.com:joergkappes/opengm-learning
i am so lazy, i do not write proper commit messages
bla
Merge branch 'master' of github.com:joergkappes/opengm-learning
improved lunary to remove number of parameters
improved lunary
changed loss interface
removed wrapped functions which will not work anyway...
removed unused header
changed loss interface (+2 squashed commits)
lunary
lunary
fubar
new example
more learners can be used
Merge branch 'master' of github.com:joergkappes/opengm-learning
mychanges
Merge branch 'master' of github.com:joergkappes/opengm-learning
changes
Merge branch 'master' of github.com:joergkappes/opengm-learning
fixed bug in gridesearchlearning
batch function gen for lpotts
batch function gen for lpotts
batch function gen for lpotts
-impmeneted batch adding of python learnable functions, NICE AS FUCK!
started to add rebing to all inference methods
fixed class vs struct bug
implemnted rebind for for almost all inference methods
implemented structured perceptron
loss graphical model type can now be specified in the dataset but has a meaningfull default templatization
better struct. perceptron
better struct. perceptron
implemented subgradient ssvm
fixed bugs...
new dataset, way
fixed minor bugs in examples
fixed minor bugs in examples
fixed stupid bugs
fixed stupid bugs
new stuff
fixed major bug in editabledataset.hxx
fixed bug i introduced
my local changes:
merged
added new dataset to python
minor stuff
added missing file
bazinga
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning
bazinga
added isPotts overload to potts function
minor changes
bazinga
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning
implemented py wrapper for maxlh-learning
fubar
added weight constraints and weight regularizer to dataset
superpixel dataset creator
minor improvements
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning
temp. disabled learn method from ml-learning in python binding
temp. disabled learn method from ml-learning in python binding
temp. disabled learn method from ml-learning in python binding
removed weight regularzier from ds (stupid idead of me to put it there)
made pyMh-learn build again
fixed h5 bugs
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning
minor changes
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning
learning changes
seems to be working
minor fix
removed deprecated function
most certainly fixed bugs
max. lh learning test less time consuming
outsmarting travis
outsmarting travis
update struct. perceptron
update struct. perceptron
update struct. perceptron
fixed that shit
fixed that shit
fixed that shit
defined CI to detect travis
give travis another shot
udpated readme
changed travis script
updated travis
updated travis
removed learning
removed lfunctions
removed py learning
removed learning wip
fixed fusion for cc
fixed include guards and WITH_*** problems for intersection based
fixing travis wip
fixing travis wip
trying hdf5 serial version
added different boost sources in travis
chaged find hdf5 script
Francesco Biscani (2):
Tentative fix for a memory leak in fast_sequence.hxx.
Disable a check in the test that results in an out-of-bounds access.
Ghislain Antony Vaillant (7):
Merge tag 'v2.3.6+20160814' into debian/master
Update patch queue.
Bump standards version to 3.9.8, no changes required.
Drop build of examples.
Skip usage of RPATH with CMake.
Assorted enhancements to rules file:
Release to unstable.
Jan Funke (27):
replaced REDME.md content with TODO list
added test-learning, added scaffolds for bundle-optimizer and struct-max-margin, fixed compile errors in hammingloss and testdataset
checkin of linear and quadratic solver and Gurobi backend for learning
bundle-optimizer: added setup of QP
added Oracle stub to opengm/learning
fixed operator[] for opengm::Weights
finished bundle method implementation
added *.swp to .gitignore
renamed opengm::{Paramters -> Weights} in bundle method
finished implementation (not testing) of StructMaxMargin
fixed gradient computation in struct-max-margin, re-enabled test
use trws in test-learning
embarrassing bug fix in bundle method
Merge remote-tracking branch 'opengm/master'
bugfix: iterator was not dereferenced in sum_of_experts
bugfix: gradient-accumulator needs to pass local configuration to function
added embarrassingly simple dataset
updated test_learning
fixed testdatasets
removed debug code from gradient-accumulator
fixed test_gridsearch_learner
re-enabled all testdatasets in test_learning
bugfix: GradientAccumulator is not supposed to initialize gradient with zero
use multiple models in test_learning
struct-max-margin: set data-weights only once for all models
added debug output to bundle-optimizer
added EpsStrategy parameter to bundle method
Janez Ales (26):
maximum likelihood files
New version for review
With likelihood function.
Change in eta.
Latest test.
No boundaries on the parameter, start at origin.
renamed file: pyMaxLikelihoodLearner.cxx
WeightGradientFunctor global weight indexing (error) changed to local weight indexing.
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning into maximum_likelihood
Belief Propagation parameters added to learner.
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning into maximum_likelihood
All parameters added.
isActive added to the belief propagation parameters
RebindGm added to LPCplex2
lweighted sum of functions calls operator() on Marray with proper indices for dimension =2 (and 1)
corrects indexing call from a learnable function.weightGradient to a multidimensional marray (*begin ---> begin)
add model print out in struct-max-margin iteration
lweightedsum_of_functions OPENGM_ASSERT corrected
adding normalized weights print output (only for check of convergence - not to be used in calculation)
paralellization added to struct-max-margin (triggered by WITH_OPENMP)
removed redundant print info
Merge branch 'learning-experimental' of https://github.com/ilastikdev/opengm-1 into structured_learning_2
verbose=false implemented in more detail
Merge branch 'learning-experimental' of https://github.com/opengm/opengm into structured_learning_2
Merge branch 'master' of https://github.com/opengm/opengm into structured_learning_2
commented out code removed
Jeffrey Yunes (2):
Added logsumexp op for log semiring (to work in log space for sum product alg)
Python bindings for log semiring
Joerg Kappes (57):
bugfixes and empty test for learnable functions
add test and learnable potts function
Merge branch 'master' of https://github.com/joergkappes/opengm-learning
start refactoring dataset
Merge branch 'master' of https://github.com/joergkappes/opengm-learning
Merge branch 'master' of https://github.com/joergkappes/opengm-learning
now models with loss are created too
Merge branch 'master' of https://github.com/joergkappes/opengm-learning
start refactoring datasets
reduce default testset size
change sign of loss in lossargumented model
adopt test to change in addloss
move test
modify cmakefiles after move
add missing resize
bugfix datasets
Merge branch 'master' of https://github.com/joergkappes/opengm-learning
add new acyclic testdataset
Update lweightedsum_of_functions.hxx
Merge pull request #374 from chaubold/fix-lpcplex2-rebindgm
Merge pull request #375 from chaubold/fix-solver-check-SSVM
Merge pull request #376 from chaubold/fix-lpgurobi-rebindgm
Merge pull request #378 from chaubold/fix-lunary-uint16
Update README.md
Update README.md
Update README.md
Merge pull request #400 from ilastikdev/structured_learning_2
Merge pull request #104 from opengm/master
fix problems with VS 2010
Merge pull request #422 from ilastikdev/structured_learning_2
Merge pull request #425 from chaubold/fix-header-patterns-when-installing
Merge pull request #442 from svenpeter42/master
disable mem-logging by default, because it might not work on all systems
Merge pull request #438 from bluescarni/partitions_test_fix
add libhdf5 to travis
Merge branch 'master' of https://github.com/opengm/opengm
add libhdf5-serial-dev to travis
add hdf5 for osx to travis
add hdf5 for osx to travis
add hdf5 for osx to travis (next try)
fix wrong namespace
Merge pull request #105 from opengm/master
Merge pull request #450 from joergkappes/master
disable python on osx for travis
Merge pull request #436 from bluescarni/memory_leak_fix
Update README.md
Merge pull request #106 from opengm/master
merge changes on inference methods from learning branch
merge changes on inference methods from learning branch
some python fixes
Merge pull request #467 from joergkappes/master
merge changes for functions and graphical model from the learning branch
merge changes for functions and graphical model from the learning branch
Merge pull request #470 from joergkappes/master
merging changes from learning-branch into master
Merge pull request #471 from joergkappes/master
Merge pull request #469 from yunesj/log_semiring
Steffen-Wolf (15):
add: unit test for dataset structure
Merge branch 'master' of github.com:joergkappes/opengm-learning
add: test for max margin learning that reproduces
make arguments of pushBackInstance in EditableDataset const
Merge branch 'master' of github.com:joergkappes/opengm-learning
fixed: resize count_ and isCached_ vector for EditableDataset
Merge branch 'master' of github.com:joergkappes/opengm-learning
add: new test to test_learning
Merge branch 'master' of github.com:joergkappes/opengm-learning
fix: added lossParams_ to editabletestdataset constructor
Merge branch 'master' of github.com:joergkappes/opengm-learning
Merge branch 'master' of github.com:joergkappes/opengm-learning
Merge branch 'master' of github.com:joergkappes/opengm-learning
Merge branch 'master' of github.com:joergkappes/opengm-learning
fix: change SumOfExperts to LSumOfExperts
Sven Peter (2):
undo dirty pull request revert hack
Revert "Merge pull request #423 from opengm/learning-experimental"
Thorsten B (11):
Merge pull request #9 from opengm/master
Merge pull request #419 from consti123/master
Merge pull request #423 from opengm/learning-experimental
Merge pull request #10 from opengm/master
Merge pull request #413 from DerThorsten/master
Merge pull request #426 from DerThorsten/master
Merge pull request #427 from DerThorsten/master
Merge pull request #428 from DerThorsten/master
Merge pull request #429 from DerThorsten/master
Merge pull request #430 from DerThorsten/master
Merge pull request #431 from DerThorsten/master
Thorsten Beier (3):
hotfix
huthut
Revert "huthut"
joergkappes (51):
prototype for learnable functions with features
add I/O for LPotts
add LPotts IO-test and do bugfixes
add directory structure for learning
typos
add interface for learnable functions
start hammingloss
prototypical implementation for loss and learning, befor adding unittests some interfaces need to be specified
todos
add dumydataset - untested
syntax checking and bugfixes
add typedef and make it larger
add missing return value
now this solver do grid search learning (todo: rename file)
add couts
write test for grid search learning(todo: rename this file)
formating
renaming
comments
Merge branch 'master' of https://github.com/joergkappes/opengm-learning
add second test-dataset (with 3 parameters) and add it to greidsearch test
change noise in datasets
save and load datasets for learning
add missing file for dataset-io
make it compileable
fix bug in deserialization of lpotts
comment out test that does not stop
change to new testmodels
fix CMameFile for test
bugfix
update cmakelists
add output
first bugfixes for maxlikelihoodlearning - need more
remove code from mll
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning
refactor WeightGradientFunctor
reimplement mll
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning
change to reimplemeted mll - test-learning and test-subgradient-ssvm fails
update mll
comment unet test out - jan/Thorsten plz check this
add temperature to mll
reduce test to shorter time
add second dataset to mll test
Typo in cmakelist
some minor fixes
Merge branch 'master' of https://bitbucket.org/jkappes/opengm-learning
fix parameter rebind
add Multicut
add learning code (experimental)
fix ambiguities in names
mschiegg (25):
rename argument, add missing include
rename gt to gts and add another dataset constructor
add editable dataset and its python wrappers
introduce default LossParameter
remove unnecessary include
bring dataset IO to python and link against HDF5
changing id of sum_of_experts
adding more safeguards
sumOfExperts serialization still to be debugged
Merge branch 'master' of github.com:joergkappes/opengm-learning
added loss serialization; test-dataset-io still fails for sum of experts
sumOfBi...Experts serialization fixed
breaks all tests. Revert "removed wrapped functions which will not work anyway..."
breaks all tests. Revert "changed loss interface"
remove obsolete test datasets
remove obsolete learnable functions
renamed sum_of_experts to lsum_of_experts for consistency
make weights_ mutable to be able to deserialize with function functor
Merge branch 'master' of github.com:joergkappes/opengm-learning
colon missing
implemented getLoss for dataset and almost brought it to python, still to be fixed: pyDataset
python scripts to convert the pystruct pascal VOC dataset to openGM format and learn
fix ifdef
fixed python wrapping, with @Steffen-Wolf
fixed loss return
opengm (2):
Update README.md
Merge pull request #399 from ilastikdev/structured_learning_2
-----------------------------------------------------------------------
No new revisions were added by this update.
--
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/debian-science/packages/opengm.git
More information about the debian-science-commits
mailing list