We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: A Generalized Hierarchical Nonnegative Tensor Decomposition

Abstract: Nonnegative matrix factorization (NMF) has found many applications including topic modeling and document analysis. Hierarchical NMF (HNMF) variants are able to learn topics at various levels of granularity and illustrate their hierarchical relationship. Recently, nonnegative tensor factorization (NTF) methods have been applied in a similar fashion in order to handle data sets with complex, multi-modal structure. Hierarchical NTF (HNTF) methods have been proposed, however these methods do not naturally generalize their matrix-based counterparts. Here, we propose a new HNTF model which directly generalizes a HNMF model special case, and provide a supervised extension. We also provide a multiplicative updates training method for this model. Our experimental results show that this model more naturally illuminates the topic hierarchy than previous HNMF and HNTF methods.
Comments: 6 pages, 2 figues, 3 tables
Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as: arXiv:2109.14820 [cs.LG]
  (or arXiv:2109.14820v2 [cs.LG] for this version)

Submission history

From: Joshua Vendrow [view email]
[v1] Thu, 30 Sep 2021 03:00:41 GMT (844kb,D)
[v2] Tue, 15 Feb 2022 18:05:00 GMT (844kb,D)

Link back to: arXiv, form interface, contact.