File python-pomegranate.changes of Package python-pomegranate
-------------------------------------------------------------------
Mon Dec 2 06:59:23 UTC 2024 - Steve Kowalik <steven.kowalik@suse.com>
- Update to 1.1.0:
* General
+ The entire codebase has been rewritten from the ground up to use
PyTorch in the backend.
* Features
+ Implement the Viterbi algorithm for DenseHMM and SparseHMM
+ GPU support has been added for all models and methods
+ Mixed/half precision has been added for all models and methods
+ Serialization is now handled by PyTorch
+ Missing values are now supported through torch.masked.MaskedTensor
objects
+ Prior probabilities are now supported for all relevant models and
methods and enable more comprehensive/flexible semi-supervised
learning than before
* Models
+ All distributions are now multivariate by default, supporting speedups
through batched operations
+ Factor graphs are now supported as first-class citizens
+ Hidden Markov models have been split into DenseHMM and SparseHMM models
which differ in how the transition matrix is encoded, with DenseHMM
objects being significantly faster
+ NaiveBayes models have been removed
+ MarkovNetworks models have been temporarily removed
+ Constraint graphs have been temporarily removed
- Drop devel package, we don't ship C sources any more.
- Switch to noarch as a result as well.
- Run the testsuite.
- Switch to pyproject macros.
- Skip Python 3.13 for now until PyTorch supports it.
-------------------------------------------------------------------
Fri Feb 12 14:56:03 UTC 2021 - Dirk Müller <dmueller@suse.com>
- skip python 36 from build
-------------------------------------------------------------------
Mon Jan 6 15:19:20 UTC 2020 - Todd R <toddrme2178@gmail.com>
- Update to Version 0.12.0
+ Highlights
* MarkovNetwork models have been added in and include both inference and structure learning.
* Support for Python 2 has been depricated.
* Markov network, data generator, and callback tutorials have been added in
* A robust `from_json` method has been added in to __init__.py that can deserialize JSONs from any pomegranate model.
+ MarkovNetwork
* MarkovNetwork models have been added in as a new probabilistic model.
* Loopy belief propagation inference has been added in using the FactorGraph backend
* Structure learning has been added in using Chow-Liu trees
+ BayesianNetwork
* Chow-Liu tree building has been sped up slightly, courtesy of @alexhenrie
* Chow-Liu tree building was further sped up by almost an order of magnitude
* Constraint Graphs no longer fail when passing in graphs with self loops, courtesy of @alexhenrie
+ BayesClassifier
* Updated the `from_samples` method to accept BayesianNetwork as an emission. This will build one Bayesian network for each class and use them as the emissions.
+ Distributions
* Added a warning to DiscreteDistribution when the user passes in an empty dictionary.
* Fixed the sampling procedure for JointProbabilityTables.
* GammaDistributions should have their shape issue resolved
* The documentation for BetaDistributions has been updated to specify that it is a Beta-Bernoulli distribution.
+ io
* New file added, io.py, that contains data generators that can be operated on
* Added DataGenerator, DataFrameGenerator, and a BaseGenerator class to inherit from
+ HiddenMarkovModel
* Added RandomState parameter to `from_samples` to account for randomness when building discrete models.
+ Misc
* Unneccessary calls to memset have been removed, courtesy of @alexhenrie
* Checking for missing values has been slightly refactored to be cleaner, courtesy of @mareksmid-lucid
* Include the LICENSE file in MANIFEST.in and simplify a bit, courtesy of @toddrme2178
* Added in a robust from_json method that can be used to deseralize a JSON for any pomegranate model.
+ docs
* Added io.rst to briefly describe data generators
* Added MarkovNetwork.rst to describe Markov networks
* Added links to tutorials that did not have tutorials linked to them.
+ Tutorials
* Added in a tutorial notebook for Markov networks
* Added in a tutorial notebook for data generators
* Added in a tutorial notebook for callbacks
+ CI
* Removed unit tests for Py2.7 from AppVeyor and Travis
* Added unit tests for Py3.8 to AppVeyor and Travis
- Dropped python2 support
-------------------------------------------------------------------
Mon Nov 18 17:12:24 UTC 2019 - Todd R <toddrme2178@gmail.com>
- Initial version
