OCTIS: Comparing Topic Models is Simple! A python package to optimize and evaluate topic models (accepted at EACL2021 demo track)

Python PyTorch (GPU) and NumPy (CPU)-based port of Févotte and Dobigeon's robust-NMF algorithm appearing in "Nonlinear hyperspectral unmixing with robust nonnegative matrix factorization."

Non-negative Matrix Factorization (NMF) Tensorflow Implementation

Optimization and Regularization variants of Non-negative Matrix Factorization (NMF)

FactorizePhys: Matrix Factorization for Multidimensional Attention in Remote Physiological Sensing [NeurIPS 2024]

An algorithm for unsupervised discovery of sequential structure

Tensor Extraction of Latent Features (T-ELF). Within T-ELF's arsenal are non-negative matrix and tensor factorization solutions, equipped with automatic model determination (also known as the estimation of latent factors - rank) for accurate data modeling. Our software suite encompasses cutting-edge data pre-processing and post-processing modules.

A blind source separation package using non-negative matrix factorization and non-negative ICA

Coupled clustering of single cell genomic data

Topic modeling streamlit app.

PyTorch implementations of the beta divergence loss.

An official implementation of "Joint Inference of Diffusion and Structure in Partially Observed Social Networks Using Coupled Matrix Factorization"

Multi-Modal Multi-Task Remote Physiological Sensing

New Matrix Factorization Algorithms based on Bregman Proximal Gradient: BPG-MF, CoCaIn BPG-MF, BPG-MF-WB

Non-Negative Matrix Factorization

Non-negative matrix factorization is applied for classification of defects on steel surface using CNN

ALPINE is a semi-supervised non-negative matrix factorization (NMF) framework designed to effectively distinguish between multiple phenotypic conditions based on shared biological factors, while also providing direct interpretability of condition-associated genes. The preprint is available on bioRxiv.

Co-clustering algorithms can seek homogeneous sub-matrices into a dyadic data matrix, such as a document-word matrix.

Filling in missing values of Sea Surface Temperature

The project develops an application that suggests to the reader more similar articles to that he already read. It uses the embedding algorithms of headlines to create their own numerical representation, which allows to compute the similarity between headlines and get the most similar ones.