Explicit Kernel and Sparsity of Eigen Subspace for the AR(1) Process

Mustafa U. Torun, Onur Yilmaz, Ali Akansu

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

Fast implementation of Karhunen-Loeve Transform (KLT) is of great interest to several disciplines, and there were attempts to derive closed-form kernel expressions for certain classes of stochastic processes. Random processes and information sources are described by stochasticsignal models, including autoregressive (AR), moving average (MA), and autoregressivemoving average (ARMA) types. This chapter focuses on the discrete autoregressive order one, AR(1), and the process and derivation of its explicit eigen kernel. It investigates the sparsity of eigen subspace and presents a rate-distortion theory-based sparsing method. The chapter then focuses on eigen subspace of a discrete AR(1) process with closed-form expressions for its eigenvectors and eigenvalues. It provides a comparative performance of the presented method along with the various methods reported in the literature. Finally, the chapter highlights the merit of the method for the AR(1) process as well as for the empirical correlation matrix of stock returns in the NASDAQ-100 index.

Original languageEnglish (US)
Title of host publicationFinancial Signal Processing and Machine Learning
PublisherWiley-IEEE Press
Pages67-99
Number of pages33
ISBN (Electronic)9781118745540
ISBN (Print)9781118745670
DOIs
StatePublished - Apr 29 2016

Fingerprint

Random processes
Eigenvalues and eigenfunctions
Mathematical transformations

All Science Journal Classification (ASJC) codes

  • Engineering(all)
  • Computer Science(all)

Keywords

  • Discrete autoregressive
  • Eigen subspace sparsity
  • Empirical correlation matrix
  • Explicit eigen kernel
  • Karhunen-Loeve transform
  • Rate-distortion theory

Cite this

Torun, M. U., Yilmaz, O., & Akansu, A. (2016). Explicit Kernel and Sparsity of Eigen Subspace for the AR(1) Process. In Financial Signal Processing and Machine Learning (pp. 67-99). Wiley-IEEE Press. https://doi.org/10.1002/9781118745540.ch5
Torun, Mustafa U. ; Yilmaz, Onur ; Akansu, Ali. / Explicit Kernel and Sparsity of Eigen Subspace for the AR(1) Process. Financial Signal Processing and Machine Learning. Wiley-IEEE Press, 2016. pp. 67-99
@inbook{aeafac654334441d8bfad08095343a18,
title = "Explicit Kernel and Sparsity of Eigen Subspace for the AR(1) Process",
abstract = "Fast implementation of Karhunen-Loeve Transform (KLT) is of great interest to several disciplines, and there were attempts to derive closed-form kernel expressions for certain classes of stochastic processes. Random processes and information sources are described by stochasticsignal models, including autoregressive (AR), moving average (MA), and autoregressivemoving average (ARMA) types. This chapter focuses on the discrete autoregressive order one, AR(1), and the process and derivation of its explicit eigen kernel. It investigates the sparsity of eigen subspace and presents a rate-distortion theory-based sparsing method. The chapter then focuses on eigen subspace of a discrete AR(1) process with closed-form expressions for its eigenvectors and eigenvalues. It provides a comparative performance of the presented method along with the various methods reported in the literature. Finally, the chapter highlights the merit of the method for the AR(1) process as well as for the empirical correlation matrix of stock returns in the NASDAQ-100 index.",
keywords = "Discrete autoregressive, Eigen subspace sparsity, Empirical correlation matrix, Explicit eigen kernel, Karhunen-Loeve transform, Rate-distortion theory",
author = "Torun, {Mustafa U.} and Onur Yilmaz and Ali Akansu",
year = "2016",
month = "4",
day = "29",
doi = "https://doi.org/10.1002/9781118745540.ch5",
language = "English (US)",
isbn = "9781118745670",
pages = "67--99",
booktitle = "Financial Signal Processing and Machine Learning",
publisher = "Wiley-IEEE Press",

}

Torun, MU, Yilmaz, O & Akansu, A 2016, Explicit Kernel and Sparsity of Eigen Subspace for the AR(1) Process. in Financial Signal Processing and Machine Learning. Wiley-IEEE Press, pp. 67-99. https://doi.org/10.1002/9781118745540.ch5

Explicit Kernel and Sparsity of Eigen Subspace for the AR(1) Process. / Torun, Mustafa U.; Yilmaz, Onur; Akansu, Ali.

Financial Signal Processing and Machine Learning. Wiley-IEEE Press, 2016. p. 67-99.

Research output: Chapter in Book/Report/Conference proceedingChapter

TY - CHAP

T1 - Explicit Kernel and Sparsity of Eigen Subspace for the AR(1) Process

AU - Torun, Mustafa U.

AU - Yilmaz, Onur

AU - Akansu, Ali

PY - 2016/4/29

Y1 - 2016/4/29

N2 - Fast implementation of Karhunen-Loeve Transform (KLT) is of great interest to several disciplines, and there were attempts to derive closed-form kernel expressions for certain classes of stochastic processes. Random processes and information sources are described by stochasticsignal models, including autoregressive (AR), moving average (MA), and autoregressivemoving average (ARMA) types. This chapter focuses on the discrete autoregressive order one, AR(1), and the process and derivation of its explicit eigen kernel. It investigates the sparsity of eigen subspace and presents a rate-distortion theory-based sparsing method. The chapter then focuses on eigen subspace of a discrete AR(1) process with closed-form expressions for its eigenvectors and eigenvalues. It provides a comparative performance of the presented method along with the various methods reported in the literature. Finally, the chapter highlights the merit of the method for the AR(1) process as well as for the empirical correlation matrix of stock returns in the NASDAQ-100 index.

AB - Fast implementation of Karhunen-Loeve Transform (KLT) is of great interest to several disciplines, and there were attempts to derive closed-form kernel expressions for certain classes of stochastic processes. Random processes and information sources are described by stochasticsignal models, including autoregressive (AR), moving average (MA), and autoregressivemoving average (ARMA) types. This chapter focuses on the discrete autoregressive order one, AR(1), and the process and derivation of its explicit eigen kernel. It investigates the sparsity of eigen subspace and presents a rate-distortion theory-based sparsing method. The chapter then focuses on eigen subspace of a discrete AR(1) process with closed-form expressions for its eigenvectors and eigenvalues. It provides a comparative performance of the presented method along with the various methods reported in the literature. Finally, the chapter highlights the merit of the method for the AR(1) process as well as for the empirical correlation matrix of stock returns in the NASDAQ-100 index.

KW - Discrete autoregressive

KW - Eigen subspace sparsity

KW - Empirical correlation matrix

KW - Explicit eigen kernel

KW - Karhunen-Loeve transform

KW - Rate-distortion theory

UR - http://www.scopus.com/inward/record.url?scp=85026434662&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85026434662&partnerID=8YFLogxK

U2 - https://doi.org/10.1002/9781118745540.ch5

DO - https://doi.org/10.1002/9781118745540.ch5

M3 - Chapter

SN - 9781118745670

SP - 67

EP - 99

BT - Financial Signal Processing and Machine Learning

PB - Wiley-IEEE Press

ER -

Torun MU, Yilmaz O, Akansu A. Explicit Kernel and Sparsity of Eigen Subspace for the AR(1) Process. In Financial Signal Processing and Machine Learning. Wiley-IEEE Press. 2016. p. 67-99 https://doi.org/10.1002/9781118745540.ch5