We propose several new models for semi-supervised nonnegative matrix factorization (SSNMF) and provide motivation for SSNMF models as maximum likelihood estimators given specific distributions of uncertainty. We present multiplicative updates training methods for each new model, and demonstrate the application of these models to classification, although they are flexible to other supervised learning tasks. We illustrate the promise of these models and training methods on both synthetic and real data, and achieve high classification accuracy on the 20 Newsgroups dataset.
@article{arxiv.2010.07956,
title = {Semi-supervised NMF Models for Topic Modeling in Learning Tasks},
author = {Jamie Haddock and Lara Kassab and Sixian Li and Alona Kryshchenko and Rachel Grotheer and Elena Sizikova and Chuntian Wang and Thomas Merkh and R. W. M. A. Madushani and Miju Ahn and Deanna Needell and Kathryn Leonard},
journal= {arXiv preprint arXiv:2010.07956},
year = {2020}
}