2.0.2

all

• Set the required version of numo-narray-alt to 0.9.10 or higher.
• Change require statement to explicitly load numo-narray-alt.

rumale-clustering and rumale-metric_learning

• Set the required version of numo-optimize to 0.2.2 or higher.

2.0.1

rumale-naive_bayes

• Fix SystemStackError that occurred in MultinomialNB when given large sample sizes: #51

others

• Refactor GitHub Actions workflow to reduce execution time.

2.0.0

all

• Change the runtime dependency from numo-narray to numo-narray-alt.

rumale-linear_model

• Change the runtime dependency from lbfgsb.rb to numo-optimize.

rumale-metric_learning

• Change the runtime dependency from lbfgsb.rb to numo-optimize.

Numo::NArray, a runtime dependency of Rumale, has not had a new release in three years. For the future development of Rumale, I have decided to fork Numo::NArray to continue its development, including applying bug fixes. There are no other major changes besides this update to the runtime dependency. In addition, it is recommended to use numo-linalg-alt if you are using Numo::Linalg with Rumale.

1.0.0

rumale-core

• Add csv gem to runtime dependencies for Ruby 3.4.

rumale-ensemble

• Add classifier and regressor classes based on Variable-Random Trees.
  • VRTreesClassifier
  • VRTreesRegressor

others

• No changes, or minor changes using RuboCop.

Note:
The above changes in this update would normally be version 0.30.0. However, considering the extensive development period of over five years, this release has been designated as version 1.0.0.

0.29.0

rumale-decomposition

• Add transformer class for Sparse Principal Component Analysis.
  • SparsePCA

rumale-manifold

• Add transformer class for Local Tangent Space Alignment.
  • LocalTangentSpaceAlignment

others

• No changes, or minor changes in configuration files.

0.28.1

rumale-core

• Fix nil checks for the y argument of euclidean_distance and squared_error methods: #45 and 4eb1727

rumale-manifold

• Add transformer classes for Hessian Eigenmaps.
  • HessianEigenmaps

others

• No changes, or minor changes using RuboCop.

0.28.0

rumale-tree

Breaking change

• Rewrite native exntension codes with C++.
• Reimplements stop_growing? private method in DecisionTreeRegressor with native extension.

rumae-neural_network

• Add classifier and regressor classes for Radial Basis Function (RBF) Network.
  • RBFClassifier
  • RBFRegressor
• Add classifier and regressor classes for Random Vector Functinal Link (RVFL) Network.
  • RVFLClassifier
  • RVFLRegressor

others

• No changes, minor changes in configuration files, or minor refactoring using RuboCop.

0.27.0

rumale-linear_model

• Add partial_fit method to SGDClassifier and SGDRegressor.
  • It performs 1-epoch of stochastic gradient descent. It only supports binary labels and single target variables.

rumale-tree

• Remove unnecessary array generation in native extension.

others

• No changes, or minor changes using RuboCop.

0.26.0

rumale-clustering

• Add cluster analysis class for mean-shift method.
  • MeanShift

rumale-manifold

• Add transformer classes for Loccally Linear Embedding and Laplacian Eigenmaps.
  • LocallyLinearEmbedding
  • LaplacianEigenmaps

rumale-metric_learning

• Add transformer class for Local Fisher Discriminant Analysis.
  • LocalFisherDiscriminantAnalysis

others

• No changes, or only slight changes to configuration files.

0.25.0

rumale-linear_model

Breaking change

• Add new SGDClassfier and SGDRegressor by extracting stochastic gradient descent solver from each linear model.
• Change the optimization method of ElasticNet and Lasso to use the coordinate descent algorithm.
• Change the optimization method of SVC and SVR to use the L-BFGS method.
• Change the loss function of SVC to the squared hinge loss.
• Change the loss function of SVR to the squared epsilon-insensitive loss.
• Change not to use random vector for initialization of weights.
  • From the above changes, keyword arguments such as learning_rate, decay, momentum, batch_size,
    and random_seed for LinearModel estimators have been removed.
• Fix the column and row vectors of weight matrix are reversed in LinearRegression, Ridge, and NNLS.

rumale-decomposition

• Fix missing require method to load Rumale::Utils in PCA class.
  It is needed to initialize the principal components when optimizing with fixed-point algorithm.

rumale-evaluation_measure

• Apply automatic correction for Style/ZeroLengthPredicate of RuboCop to ROCAUC class.

others

• No changes, or only modifications in test code or configuration.