Releases: ncaptier/stabilized-ica
Releases · ncaptier/stabilized-ica
PyPI release version 2.0.0
Stabilized-ica is now compatible with scikit-learn API, meaning that you can use the base class as a sklearn transformer and include it in complex ML pipelines (see this tutorial for an illustration).
sica.annotate and sica.singlecell modules have been removed from stabilized-ica and integrated into a complementary python toolbox called sica-omics . stabilized-ica no longer contains dependencies specific to omics data analysis.
Fixed bugs:
- svd_solver default value (parameter of sica._whitening.whitening) was changed from full (i.e full svd decomposition) to auto (i.e selection of most efficient solver for the size of the given dataset). This significantly speeds up the computation for large datasets.
New features:
- sica.base.MSTD has new fun and algorithm parameters so that the user can specify the ICA algorithm and the non-linearity function to use (for the previous version only algorithm = fastica_par and fun = 'logcosh' were available).
PyPI release version 1.1.0
Release highlights :
- Add bootstrap options for sica.base.StabilizedICA
- Correct some bugs
- Add new tutorials
Pre-release for v1.1.0
Add bootstrap functionality for the stabilized ICA algorithm
First release on PyPi
v1.0.1 Update setup.py
Pre-release of the whole package (for sanity check)
v1.0.0-alpha Update README.md