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


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


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

Computer Science > Machine Learning

Title: HyperNTF: A Hypergraph Regularized Nonnegative Tensor Factorization for Dimensionality Reduction

Abstract: Most methods for dimensionality reduction are based on either tensor representation or local geometry learning. However, the tensor-based methods severely rely on the assumption of global and multilinear structures in high-dimensional data; and the manifold learning methods suffer from the out-of-sample problem. In this paper, bridging the tensor decomposition and manifold learning, we propose a novel method, called Hypergraph Regularized Nonnegative Tensor Factorization (HyperNTF). HyperNTF can preserve nonnegativity in tensor factorization, and uncover the higher-order relationship among the nearest neighborhoods. Clustering analysis with HyperNTF has low computation and storage costs. The experiments on four synthetic data show a desirable property of hypergraph in uncovering the high-order correlation to unfold the curved manifolds. Moreover, the numerical experiments on six real datasets suggest that HyperNTF robustly outperforms state-of-the-art algorithms in clustering analysis.
Comments: 12 pages, 6 figures, 9 tables
Subjects: Machine Learning (cs.LG); Image and Video Processing (eess.IV); Signal Processing (eess.SP)
Cite as: arXiv:2101.06827 [cs.LG]
  (or arXiv:2101.06827v1 [cs.LG] for this version)

Submission history

From: Quanying Liu [view email]
[v1] Mon, 18 Jan 2021 01:38:47 GMT (689kb,D)
[v2] Tue, 26 Jan 2021 16:08:29 GMT (690kb,D)

Link back to: arXiv, form interface, contact.