Reproduce “A Neural Network for Determination of Latent Dimensionality in Nonnegative Matrix Factorization” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Benjamin T. Nebgen, Raviteja Vangara, Miguel A. Hombrados-Herrera, Svetlana Kuksova, Boian S. Alexandrov
The authors train an MLP classifier to select the number of latent features, K, in nonnegative matrix factorization (NMF). It reads sliding windows of NMFk reconstruction and stability statistics, then combines window predictions by voting. On 9,969 held-out synthetic matrices, it recovered the planted K exactly in 95.37% of cases and within one in 99.54%; real-data results were mixed.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.