1.7.6

new release for easier installation

What's Changed

• don't write intermediate U-matrix when s=0 by @yoch in #116
• Fix a serious bug by @yao531441 in #125
• Provisional fix for Issue #130 by @MattWenham in #131
• Add possibility to choose the order of the returned numpy array in Somoclu.get_bmus() by @giacomolanciano in #146
• src/Makefile.in: fix compilation rule for libsomoclu so that make cor… by @tomcucinotta in #150
• Update LICENSE by @achapkowski in #152
• Minor fixes by @tomcucinotta in #158
• Add norm-p as distance metric (with positive real p) by @tomcucinotta in #160

New Contributors

• @yao531441 made their first contribution in #125
• @MattWenham made their first contribution in #131
• @giacomolanciano made their first contribution in #146
• @tomcucinotta made their first contribution in #150
• @achapkowski made their first contribution in #152

Full Changelog: 1.7.5...1.7.6

1.7.5

• New: A Makefile for mingw to build on Windows.
• Changed: PR #94 added a much more efficient sparse kernel.
• Changed: boilerplate code for Julia greatly improved.
• Changed: Code cleanup, pre-processor macros simplified.
• Changed: Adapted to Seaborn API changes in plotting heatmaps.

1.7.4

• Fixed: The random seed was set to 0 for testing purposes. This is now changed to a wall-time based initialization.

1.7.3

• New: Verbosity parameter in the command-line, Python, MATLAB, and Julia interfaces.
• Changed: Calculation of U-matrix parallelized.
• Changed: Moved feeding data to train method in the Python interface.
• Fixed: Sparse matrix reader made more robust.
• Fixed: Compatibility with kohonen 3 resolved.
• Fixed: Compatibility with Matplotlib 2 resolved.

1.7.2

• New: The coefficient of the Gaussian neighborhood function exp(-||x-y||^2/(2*(coeff*radius)^2)) is now exposed in all interfaces as a parameter.
• New: get_bmu function in the Python interface to get the best matching units given an activation map.
• Changed: Updated PCA initialization in the Python interface to work with sk-learn 0.18 onwards.
• Changed: Radii can be float values.
• Fixed: Only positive values were written back to codebook during update.
• Fixed: Sparse data is read correctly when there are class labels.

1.7.1

• Fixed: macOS build works again.

1.7.0

• New: Julia interface is available (https://github.com/peterwittek/Somoclu.jl).
• New: Method get_surface_state of the Somoclu object in Python calculates the activation map for all data instances.
• New: Method view_activation_map of the Somoclu object in Python allows plotting the activation map for the training data instances or for a new data instance.
• New: Method view_similarity_matrix of the Somoclu object in Python visualizes the similarity matrix of data points according to their distance to the nodes in the map.
• Fixed: CRAN-friendliness improved.

1.6.2

• Changed: In-place codebook updates when compiled without MPI. This improves update speed and substantially cuts memory use.
• Changed: Compatible with Visual Studio 15.
• Fixed: The BMUs returned after training were from before the last epoch. Now another round of BMU search is done.
• Fixed: Training can continue on the same data in the Python wrapper.
• Fixed: GPU memory allocation problem on Windows.

1.6.1

• New: Option for PCA initialization is added to the Python interface.
• New: Clustering of the codebook with arbitrary clustering algorithm in scikit-learn is now possible in the Python interface.

1.6

• New: R wrapper integrates with kohonen package.
• New: MATLAB wrapper integrates with soomtoolbox.
• New: Better handling of CUDA compilation in the Python interface.
• Changed: Throws an exception if GPU kernel is requested, but it was compiled without it. The earlier behaviour quietly defaulted to the CPU kernel.