**RcppML** is an R package for high-performance **Non-negative Matrix Factorization** (NMF), truncated **SVD/PCA**, and divisive clustering of large sparse and dense matrices. It supports six statistical distributions via IRLS, cross-validation for automatic rank selection, optional CUDA GPU acceleration, out-of-core streaming for datasets larger than memory, and a composable graph DSL for multi-modal and deep NMF.
0 commit comments