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

Download:

Current browse context:

cs.LG

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 consistent and flexible framework for deep matrix factorizations

Abstract: Deep matrix factorizations (deep MFs) are recent unsupervised data mining techniques inspired by constrained low-rank approximations. They aim to extract complex hierarchies of features within high-dimensional datasets. Most of the loss functions proposed in the literature to evaluate the quality of deep MF models and the underlying optimization frameworks are not consistent because different losses are used at different layers. In this paper, we introduce two meaningful loss functions for deep MF and present a generic framework to solve the corresponding optimization problems. We illustrate the effectiveness of this approach through the integration of various constraints and regularizations, such as sparsity, nonnegativity and minimum-volume. The models are successfully applied on both synthetic and real data, namely for hyperspectral unmixing and extraction of facial features.
Comments: 21 pages, 21 figures
Subjects: Machine Learning (cs.LG); Signal Processing (eess.SP); Numerical Analysis (math.NA); Machine Learning (stat.ML)
Cite as: arXiv:2206.10693 [cs.LG]
  (or arXiv:2206.10693v1 [cs.LG] for this version)

Submission history

From: Nicolas Gillis [view email]
[v1] Tue, 21 Jun 2022 19:20:35 GMT (12528kb,D)

Link back to: arXiv, form interface, contact.