My experiment of multilayer NMF, a deep neural network in which the first several layers take Semi-NMF as its pseudo-activation-function that finds the latent sturcture embedding in the original data unsupervisely.

This repository contains the pytorch code for the paper: A Deep Semi-NMF Model for Learning Hidden Representations (Trigeorgis2014)

Official implementation of our NCC'24 paper titled "Semi‑NMF Regularization‑Based Autoencoder Training for Hyperspectral Unmixing".