Matrix factorization classes (stnmf.mf)

These classes are decoupled from the concepts of STNMF. The matrix factorizations follow the nomenclature and descriptors of the non-negative matrix factorization (NMF) literature. The objective is to solve

\(\mathbf{V} \approx \mathbf{W}\mathbf{H} \,,\)

where \(V \in \mathbb{R}^{n\times m}\) is the input matrix consisting of \(n\) variables and \(m\) observations, \(W \in \mathbb{R}^{n\times r}\) are the \(r\) recovered features and \(H \in \mathbb{R}^{r\times m}\) are the corresponding encodings.

The classes inherit from the class MF that decomposes \(V\) through iterative updates of \(W\) and \(H\) to be implemented by the sub classes.

MF(v, r[, w0, seed, rng, dtype])	Abstract matrix factorization class
SemiNMF(v, r[, w0, seed, rng, dtype])	Abstract semi-non-negative matrix factorization (NMF) class
SparseSemiNMF(*args[, sparsity])	Sparse semi-non-negative matrix factorization (NMF)