Tensor sparse coding for positive definite matrices

Ravishankar Sivalingam, Daniel L Boley, Vassilios Morellas, Nikolaos P Papanikolopoulos

Research output: Contribution to journalArticle

19 Citations (Scopus)

Abstract

In recent years, there has been extensive research on sparse representation of vector-valued signals. In the matrix case, the data points are merely vectorized and treated as vectors thereafter (for example, image patches). However, this approach cannot be used for all matrices, as it may destroy the inherent structure of the data. Symmetric positive definite (SPD) matrices constitute one such class of signals, where their implicit structure of positive eigenvalues is lost upon vectorization. This paper proposes a novel sparse coding technique for positive definite matrices, which respects the structure of the Riemannian manifold and preserves the positivity of their eigenvalues, without resorting to vectorization. Synthetic and real-world computer vision experiments with region covariance descriptors demonstrate the need for and the applicability of the new sparse coding model. This work serves to bridge the gap between the sparse modeling paradigm and the space of positive definite matrices.

Original languageEnglish (US)
Article number6574845
Pages (from-to)592-605
Number of pages14
JournalIEEE Transactions on Pattern Analysis and Machine Intelligence
Volume36
Issue number3
DOIs
StatePublished - Mar 1 2014

Fingerprint

Sparse Coding
Positive definite matrix
Tensors
Vectorization
Tensor
Eigenvalue
Symmetric Positive Definite Matrix
Sparse Representation
Positivity
Computer Vision
Descriptors
Patch
Riemannian Manifold
Paradigm
Computer vision
Modeling
Demonstrate
Experiment
Model
Experiments

Keywords

  • Sparse coding
  • computer vision
  • optimization
  • positive definite matrices
  • region covariance descriptors

Cite this

Tensor sparse coding for positive definite matrices. / Sivalingam, Ravishankar; Boley, Daniel L; Morellas, Vassilios; Papanikolopoulos, Nikolaos P.

In: IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 36, No. 3, 6574845, 01.03.2014, p. 592-605.

Research output: Contribution to journalArticle

@article{54b956cc552144ec81512910ab70661f,
title = "Tensor sparse coding for positive definite matrices",
abstract = "In recent years, there has been extensive research on sparse representation of vector-valued signals. In the matrix case, the data points are merely vectorized and treated as vectors thereafter (for example, image patches). However, this approach cannot be used for all matrices, as it may destroy the inherent structure of the data. Symmetric positive definite (SPD) matrices constitute one such class of signals, where their implicit structure of positive eigenvalues is lost upon vectorization. This paper proposes a novel sparse coding technique for positive definite matrices, which respects the structure of the Riemannian manifold and preserves the positivity of their eigenvalues, without resorting to vectorization. Synthetic and real-world computer vision experiments with region covariance descriptors demonstrate the need for and the applicability of the new sparse coding model. This work serves to bridge the gap between the sparse modeling paradigm and the space of positive definite matrices.",
keywords = "Sparse coding, computer vision, optimization, positive definite matrices, region covariance descriptors",
author = "Ravishankar Sivalingam and Boley, {Daniel L} and Vassilios Morellas and Papanikolopoulos, {Nikolaos P}",
year = "2014",
month = "3",
day = "1",
doi = "10.1109/TPAMI.2013.143",
language = "English (US)",
volume = "36",
pages = "592--605",
journal = "IEEE Transactions on Pattern Analysis and Machine Intelligence",
issn = "0162-8828",
publisher = "IEEE Computer Society",
number = "3",

}

TY - JOUR

T1 - Tensor sparse coding for positive definite matrices

AU - Sivalingam, Ravishankar

AU - Boley, Daniel L

AU - Morellas, Vassilios

AU - Papanikolopoulos, Nikolaos P

PY - 2014/3/1

Y1 - 2014/3/1

N2 - In recent years, there has been extensive research on sparse representation of vector-valued signals. In the matrix case, the data points are merely vectorized and treated as vectors thereafter (for example, image patches). However, this approach cannot be used for all matrices, as it may destroy the inherent structure of the data. Symmetric positive definite (SPD) matrices constitute one such class of signals, where their implicit structure of positive eigenvalues is lost upon vectorization. This paper proposes a novel sparse coding technique for positive definite matrices, which respects the structure of the Riemannian manifold and preserves the positivity of their eigenvalues, without resorting to vectorization. Synthetic and real-world computer vision experiments with region covariance descriptors demonstrate the need for and the applicability of the new sparse coding model. This work serves to bridge the gap between the sparse modeling paradigm and the space of positive definite matrices.

AB - In recent years, there has been extensive research on sparse representation of vector-valued signals. In the matrix case, the data points are merely vectorized and treated as vectors thereafter (for example, image patches). However, this approach cannot be used for all matrices, as it may destroy the inherent structure of the data. Symmetric positive definite (SPD) matrices constitute one such class of signals, where their implicit structure of positive eigenvalues is lost upon vectorization. This paper proposes a novel sparse coding technique for positive definite matrices, which respects the structure of the Riemannian manifold and preserves the positivity of their eigenvalues, without resorting to vectorization. Synthetic and real-world computer vision experiments with region covariance descriptors demonstrate the need for and the applicability of the new sparse coding model. This work serves to bridge the gap between the sparse modeling paradigm and the space of positive definite matrices.

KW - Sparse coding

KW - computer vision

KW - optimization

KW - positive definite matrices

KW - region covariance descriptors

UR - http://www.scopus.com/inward/record.url?scp=84894571127&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84894571127&partnerID=8YFLogxK

U2 - 10.1109/TPAMI.2013.143

DO - 10.1109/TPAMI.2013.143

M3 - Article

VL - 36

SP - 592

EP - 605

JO - IEEE Transactions on Pattern Analysis and Machine Intelligence

JF - IEEE Transactions on Pattern Analysis and Machine Intelligence

SN - 0162-8828

IS - 3

M1 - 6574845

ER -