cebmf_torch
cebmf_torch: Empirical Bayes Matrix Factorization in PyTorch¶
Authors: William Denault
A pure-PyTorch implementation of Empirical Bayes Matrix Factorization (EBMF) and Empirical Bayes Normal Means (EBNM) methods.
It is designed for scalable, GPU-accelerated analysis of large datasets, with a focus on genomics and other high-dimensional applications. The package provides flexible prior families, efficient mini-batch EM, and full support for GPU computation.