denoiseR


The denoiseR package is dedicated to estimate a low rank matrix from noisy data. Different regularized versions of the singular values decomposition (SVD) are implemented: soft­-thresholding of the singular values, non­linear transformation, adaptive shrinkage, etc. Iterative Stable Autoencoder is also implemented and is a new way to denoise count data.

[denoiseR website] – [CRAN package]