All noteable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
- Refactor using the AnnData and MuData data structures
- Support fixing the model variance of (multimodal) CorrNMF models during training
- Improve signature, history and embedding plots
- Support a sample-weigted KL-divergence loss in KL-NMF
- Support a sample-weighted sparsity regularization in KL-NMF
- Support fixing signature and sample biases of (multimodal) CorrNMF models during training
- Improve CorrNMF model formulation (added signature biases)
- Improve multimodal exposure plot
- Support fixing arbitrary many a priori known signatures during model training
- Improved performance with just-in-time compiled update rules
- First release of the non-negative matrix factorization (NMF) framework. Implemented algorithms: NMF with the generalized Kullback-Leibler divergence (KL-NMF), minimum-volume NMF (mvNMF), a version of correlated NMF (CorrNMF), a multimodal version of correlated NMF (MultimodalCorrNMF).
- Install:
pip install salamander-learn==0.1.0